An Official Journal of the American Academy of Neurology Neurology.org/nn○Online ISSN: 2332-7812 Volume 4, Number 5, September 2017 Neuroimmunology & Neuroinfl ammation

IgLON5 antibody: MRI evaluation of α4-integrin receptor Neurological thalamic volume desaturation and disease accompaniments & differentiates multiple activity return after outcomes in 20 patients sclerosis from common natalizumab cessation mimics Table of Contents Neurology.org/nn  Online ISSN: 2332-7812 Volume 4, Number 5, September 2017

EDITOR’S CORNER e381 High-dose cyclophosphamide without stem cell e394 Shifting paradigms rescue in immune-mediated necrotizing J. Dalmau myopathies C.A. Mecoli, A.H. Lahouti, R.A. Brodsky, A.L. Mammen, and L. Christopher-Stine EDITORIAL e393 Understanding anti-IgLON5 disease – F. Graus and J. Santamaría e382 Varicella zoster virus infected cerebrovascular cells Companion articles e385, e383 produce a proinflammatory environment D. Jones, C.P. Neff, B.E. Palmer, K. Stenmark, and M.A. Nagel ARTICLES e385 IgLON5 antibody: Neurological accompaniments and outcomes in 20 patients e387 MRI evaluation of thalamic volume differentiates MS J.A. Honorat, L. Komorowski, K.A. Josephs, K. Fechner, from common mimics E.K. St Louis, S.R. Hinson, S. Lederer, N. Kumar, A.J. Solomon, R. Watts, B.E. Dewey, and D.S. Reich A. Gadoth, V.A. Lennon, S.J. Pittock, and A. McKeon Editorial e393, Companion article e383 e388 a4-integrin receptor desaturation and disease activity e375 Reduced rich-club connectivity is related to disability return after natalizumab cessation in primary progressive MS T. Derfuss, J.M. Kovarik, L. Kappos, M. Savelieva, J.-P. Stellmann, S. Hodecker, B. Cheng, N. Wanke, R. Chhabra, A. Thakur, Y. Zhang, H. Wiendl, and K.L. Young, C. Hilgetag, C. Gerloff, C. Heesen, D. Tomic G. Thomalla, and S. Siemonsen

e390 Teriflunomide slows BVL in relapsing MS: A reanalysis e374 Lipoic acid in secondary progressive MS: A randomized of the TEMSO MRI data set using SIENA controlled pilot trial E.-W. Radue, T. Sprenger, L. Gaetano, R. Spain, K. Powers, C. Murchison, E. Heriza, K. Winges, N. Mueller-Lenke, S. Cavalier, K. Thangavelu, V. Yadav, M. Cameron, E. Kim, F. Horak, J. Simon, and M.A. Panzara, J.E. Donaldson, F.M. Woodward, D. Bourdette J. Wuerfel, J.S. Wolinsky, and L. Kappos

e378 Glucocorticoid-associated blood glucose response VIEWS & REVIEWS and MS relapse recovery e386 Frequencies of neuronal autoantibodies in healthy M.D. Goldman, S. Koenig, C. Engel, C.R. McCartney, controls: Estimation of disease specificity and M.-W. Sohn K. Lang and H. Prüss

e379 Predicting long-term disability outcomes in patients CLINICAL/SCIENTIFIC NOTES with MS treated with teriflunomide in TEMSO e383 IgLON5 disease responsive to immunotherapy M.P. Sormani, P. Truffinet, K. Thangavelu, P. Rufi, M. Bonello, A. Jacob, M.A. Ellul, E. Barker, R. Parker, C. Simonson, and N. De Stefano S. Jefferson, and S. Alusi Editorial e393, Companion article e385 Table of Contents continued e377 Dramatic rebounds of MS during pregnancy following e384 Natalizumab granule cell neuronopathy: FDG-PET in fingolimod withdrawal diagnosis and immune reconstitution with G-CSF G. Novi, A. Ghezzi, M. Pizzorno, C. Lapucci, S. Kamourieh, K. Gananandan, J. Raffel, and R. Nicholas F. Bandini, P. Annovazzi, G.L. Mancardi, and A. Uccelli e389 Long-term improvement after combined immunomodulation in early post-H1N1 vaccination narcolepsy e376 Opsoclonus-myoclonus syndrome during rituximab R. Viste, J. Soosai, T. Vikin, P.M. Thorsby, K.B. Nilsen, treatment for autoimmune autonomic and S. Knudsen ganglionopathy O.M. Dumitrascu, A. McKeon, L. Zuniga, M.F. Grill, and B.P. Goodman e392 Postmalaria neurologic syndrome associated with neurexin-3a antibodies A. Costa, A. Silva-Pinto, J. Alves, N. Neves, E. Martínez-Hernández, P. Abreu, and A. Sarmento e380 Lipoic acid pharmacokinetics at baseline and 1 year in secondary progressive MS F. Bittner, C. Murchison, D. Koop, D. Bourdette, and CORRECTION R. Spain e391 Both cladribine and alemtuzumab may effect MS via B-cell depletion

Video LOE classification LOE recommendation Podcast

tinyurl.com/NeurologyNN twitter.com/GreenJournal

Cover image: Schematic representation of the rich-club organization in average primary-progressive multiple sclerosis connectome. Stylized by Andrea Rahkola, Neurology Production Editor. See “Reduced rich-club connectivity is related to disability in primary-progressive MS.” Neurology.org/nn  Online ISSN: 2332-7812 Volume 4, Number 5, September 2017

fi VISION: Neurology® Neuroimmunology & Neuroinflammation will be Academy Of cers the premier peer-reviewed journal for experts in the fields of Ralph L. Sacco, MD, MS, FAAN, President fl James C. Stevens, MD, FAAN, President Elect Neuroimmunology and Neuroin ammation. Ann H. Tilton, MD, FAAN, Vice President Carlayne E. Jackson, MD, FAAN, Secretary MISSION: Neurology: Neuroimmunology & Neuroinflammation will Janis M. Miyasaki, MD, MEd, FRCPC, FAAN, Treasurer provide neurologists with peer-reviewed articles, editorials, and Terrence L. Cascino, MD, FAAN, Past President reviews to enhance patient care, education, and clinical & Executive Office, American Academy of Neurology translational research. Catherine M. Rydell, CAE, Executive Director/CEO 201 Chicago Ave Editor Minneapolis, MN 55415 Neurology: Neuroimmunology & Neuroinflammation Tel: 612-928-6100 Josep O. Dalmau, MD, PhD fi ICREA-University of Barcelona Editorial Of ce Barcelona, Spain Patricia K. Baskin, MS, Executive Editor University of Pennsylvania Kathleen M. Pieper, Senior Managing Editor, Neurology Philadelphia, PA Sharon L. Quimby, Managing Editor, Neurology® Clinical Practice Tel: 612-928-6400 Morgan S. Sorenson, Managing Editor, Neurology: Neuroimmunology fl Toll-free: 800-957-3182 (US) & Neuroin ammation Fax: 612-454-2748 Lee Ann Kleffman, Managing Editor, Neurology® Genetics [email protected] Cynthia S. Abair, MA, Senior Graphics Editor Research focus: Autoimmune and paraneoplastic disorders, described Andrea R. Rahkola, Production Editor, Neurology a new category of disorders mediated by antibodies against Robert J. Witherow, Senior Editorial Associate neuronal cell surface and synaptic proteins Karen Skaja, Senior Editorial Associate Kaitlyn Aman Ramm, Editorial Assistant Kristen Swendsrud, Editorial Assistant Editor-in-Chief Andrea Willgohs, Editorial Assistant Neurology® Robert A. Gross, MD, PhD, FAAN Publisher Professor of Neurology and of Pharmacology and Physiology Wolters Kluwer Strong Epilepsy Center Baltimore, MD University of Rochester Medical Center Rochester, NY Publishing Staff Tel: 612-928-6400 Kim Jansen, Executive Publisher Toll-free: 800-957-3182 (US) Alexandra Lazerow, Production Team Leader, AAN Journals Fax: 612-454-2748 Mark Armacost, Production Editor [email protected] Stacy Drossner, Production Associate

Editorial Inquiries Tel: 612-928-6400 Toll-free: 800-957-3182 (US) Fax: 612-454-2748 [email protected] Neurology.org/nn  Online ISSN: 2332-7812 Volume 4, Number 5, September 2017

Editor Christopher A. Beck, PhD Neurology® Neuroimmunology & Neuroinflammation University of Rochester Josep O. Dalmau, MD, PhD Rochester, NY ICREA-University of Barcelona Sue Leurgans, PhD Barcelona, Spain Rush University Medical Center University of Pennsylvania Chicago, IL Philadelphia, PA Level of Evidence Evaluations Editor-in-Chief Gary S. Gronseth, MD, FAAN Neurology® University of Kansas Robert A. Gross, MD, PhD, FAAN Kansas City, KS Professor of Neurology and of Pharmacology and Physiology Strong Epilepsy Center Podcasts University of Rochester Medical Center Ted M. Burns, MD, Deputy Podcast Editor Rochester, NY University of Virginia Charlottesville, VA Deputy Editors Andrew M. Southerland, MD, MSc Neurology University of Virginia Bradford B. Worrall, MD, MSc, FAAN Charlottesville, VA Professor of Neurology and Public Health Sciences University of Virginia Ombudsman Charlottesville, VA David S. Knopman, MD, FAAN Mayo Clinic Neurology: Neuroimmunology & Neuroinflammation Rochester, MN Scott S. Zamvil, MD, PhD, FAAN Scientific Integrity Advisor Department of Neurology Robert B. Daroff, MD, FAAN University of California, San Francisco Case Western Reserve University San Francisco, CA Cleveland, OH Research focus: Human and animal models of CNS autoimmune diseases, identified mechanisms used by B cells Editorial Board Associate Editors Rohit Aggarwal, MD, MS, University of Pittsburgh, Pittsburgh, PA Neurology: Neuroimmunology & Neuroinflammation Sergio E. Baranzini, PhD, University of San Francisco, San Francisco, CA Susanne Benseler, MD, Calgary, Alberta, Canada Michael Heneka, MD Jeffrey L. Bennett, MD, PhD, FAAN, University of Colorado, Aurora, CO University of Bonn Séverine Boillée, PhD, Hôpital de la Salpêtriere, Paris, France Bonn, Germany Matthijs C. Brouwer, MD, PhD, Academic Medical Center, Amsterdam, Specialties: Neurodegeneration, stroke The Netherlands Research focus: Interplay between neurodegeneration and the immune system David B. Clifford, MD, FAAN, Washington University School of Medicine, Dennis Kolson, MD, PhD St.Louis,MO University of Pennsylvania Russell C. Dale, MBChB, MRCP, PhD, University of Sydney, Sydney, Philadelphia, PA Australia Specialty: HIV, Infectious disease Oscar H. Del Brutto, MD, FAAN, School of Medicine, Research focus: Mechanisms and determinants of HIV-induced neuronal Universidad Espiritu Santo, Samborondon, Ecuador injury Thomas G. Forsthuber, MD, University of Texas, San Antonio, TX Friedmann Paul, MD Joseph El Khoury, MD, Massachusetts General Hospital, Boston, MA Charite University Hospital Markus P. Kummer, PhD, University of Bonn, Bonn, Germany Berlin, Germany Jun Li, MD, PhD, Vanderbilt University, Nashville, TN Specialties: Multiple sclerosis, neuromyelitis optica, neuroimaging Lin Mei, MD, PhD, Georgia Regents University, Augusta, Georgia (magnetic resonance imaging, optical coherence tomography) Christopher Power, MD, University of Alberta, Alberta, Canada Research focus: CNS autoimmune disease diagnosis using novel imaging Richard M. Ransohoff, MD, Biogen Idec, Boston, MA techniques Christine Stadelmann, MD, University of Göttingen, Göttingen, Germany Israel Steiner, MD, Rabin Medical Center, Petach Tikva, Israel Lan Zhou, MD, PhD Lawrence Steinman, MD, FAAN, Stanford University, Stanford, CA Mount Sinai School of Medicine JesperTegnér,PhD,MSc,KarolinskaUniversity Hospital, Stockholm, Sweden New York, NY SilviaN.Tenembaum,MD,NationalPediatricHospital,JuanP.Garrahan, Specialty: Neuromuscular Buenos Aires, Argentina Research focus: Macrophage subsets in acute skeletal muscle injury repair Maarten J. Titulaer, MD, PhD, Erasmus Medical Center, Rotterdam, Biostatistics The Netherlands Richard J. Kryscio, PhD Ari Waisman, PhD, University of Mainz, Mainz, Germany University of Kentucky Hugh J. Willison, MBBS, PhD, University of Glasgow, Glasgow, UK Lexington, KY Gregory F. Wu, MD, PhD, Washington University in St. Louis, St. Louis, MO Neurology.org/nn  Online ISSN: 2332-7812 Volume 4, Number 5, September 2017

Level of Evidence Review Team David S. Gloss II, MD, MPH&TM, Charleston, WV Melissa J. Armstrong, MD, Gainesville, FL John J. Halperin, MD, FAAN, Summit, NJ Richard L. Barbano, MD, PhD, FAAN, Rochester, NY Jason Lazarou, MSc, MD, Toronto, Ontario, Canada Richard M. Dubinsky, MD, MPH, FAAN, Kansas City, KS Steven R. Messé, MD, FAAN, Philadelphia, PA Jeffrey J. Fletcher, MD, MSc, Ann Arbor, MI Pushpa Narayanaswami, MBBS, DM, FAAN, Boston, MA Gary M. Franklin, MD, MPH, FAAN, Seattle, WA Alex Rae-Grant, MD, Cleveland, OH

Publication Information Neurology® is a registered trademark of the American Academy of Neurology (registration valid in the United States). Neurology® Neuroimmunology & Neuroinflammation (eISSN 2332-7812) is an open access journal published online for the American Academy of Neurology, 201 Chicago Ave, Minneapolis, MN 55415, by Wolters Kluwer Health, Inc. at 14700 Citicorp Drive, Bldg 3, Hagerstown, MD 21742. Business offices are located at Two Commerce Square, 2001 Market St., Philadelphia, PA 19103. Production offices are located at 351 West Camden Street, Baltimore, MD 21201-2436. © 2017 American Academy of Neurology. Neurology® Neuroimmunology & Neuroinflammation is an official journal of the American Academy of Neurology. Journal website: Neurology.org/nn, AAN website: AAN.com Copyright and Permission Information: Please go to the journal website (www.neurology.org/nn) and click the “©Request Permissions” icon for the relevant article. Alternatively, send an email to [email protected]. General information about permissions can be found here: https://www.lww.com/journal-permission. Disclaimer: Opinions expressed by the authors and advertisers are not necessarily those of the American Academy of Neurology, its affiliates, or of the Publisher. The American Academy of Neurology, its affiliates, and the Publisher disclaim any liability to any party for the accuracy, completeness, efficacy, or availability of the material contained in this publication (including drug dosages) or for any damages arising out of the use or non-use of any of the material contained in this publication. Advertising Sales Representatives: Wolters Kluwer, 333 Seventh Avenue, New York, NY 10001. Contacts: Eileen Henry, tel: 732-778-2261, fax: 973-215-2485, [email protected] and Elizabeth S. Hall, tel: 267-804-8123, [email protected]. In Europe: Avia Potashnik, Wolters Kluwer, tel: 144 207 981 0722; 144 7919 397 933 or e-mail: [email protected]. Careers & Events: Monique McLaughlin, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: (215) 521- 8468, fax: 215-521-8801; [email protected]. Reprints: Meredith Edelman, Commercial Reprint Sales, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-555-1212 (office), 215-356-2721 (mobile); [email protected]; [email protected]. Special projects: US & Canada: Alan Moore, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-521-8638, [email protected]. International: Andrew Wible, Senior Manager, Rights, Licensing, and Partnerships, Wolters Kluwer, [email protected]. EDITOR’SCORNER Shifting paradigms

Josep Dalmau, MD, PhD The classic paradigm of antibody-associated encepha- immunoglobulins, plasma exchange, and pulses of litis is that the majority of these diseases evolve rap- cyclophosphamide). idly, with CSF or MRI abnormalities suggesting an On a different topic, Solomon et al.6 investigated Correspondence to inflammatory process and pathologic findings show- whether MRI evaluation of thalamic volume differ- Dr. Dalmau: [email protected] ing inflammatory infiltrates. Well, hold on to your entiates MS from other disorders that cause white hats because here comes anti-IgLON5 disease.1 This matter abnormalities. The question is important, Neurol Neuroimmunol remarkable disease develops with core symptoms that considering the absence of highly specific biomarkers Neuroinflamm include abnormal behaviors and vocalizations during for MS and the considerable number of diseases that 2017;4:e394; doi: 10.1212/ NXI.0000000000000394 non-REM and REM sleep, preceded or accompanied can mimic the clinical and radiographic findings of by bulbar dysfunction, gait instability, cognitive dete- this disorder. The authors focused on thalamic atro- rioration, or chorea (see the editorial comment by phy, given that it is identified early in the course of Graus and Santamaria).2 The clinical picture can MS, including pediatric MS, presymptomatic MS, evolve for several years and associates with features “radiologically isolated syndrome,” and “clinically iso- suggesting an autoimmune disorder (antibodies lated syndrome.” The study included 40 participants, against a neuronal cell surface protein [IgLON5] 10 for each of the following groups: MS without and a strong association with the HLA- comorbidities for white matter abnormalities, MS DRB1*10:01), but on the other hand, the pathologic with comorbidities for white matter abnormalities, studies show a novel neuronal tauopathy predomi- migraine with MRI white matter abnormalities, and nantly affecting the hypothalamus and brainstem.3 patients who had been incorrectly diagnosed with Moreover, different from most encephalopathies MS. The findings provide Class IV evidence that associated with neuronal cell surface antibodies, the MRI volumetric evaluation of the thalamus, but not response to immunotherapy seems to be suboptimal. other deep gray matter structures, differentiates MS Although the disease is rare (the incidence and prev- from the other disorders that cause white matter alence are still unknown), it has stirred up the interest changes and that are often mistaken for MS. The from different specialties with an increasing number authors indicate that future studies should include of single case reports and a few series. In this issue of larger number of patients along with the development Neurology® Neuroimmunology & Neuroinflammation, of automated and easily applied volumetric assess- Honorat et al.4 report their experience at Mayo ment applicable to clinical practice. Clinic, with 26 patients harboring these antibodies, In another study, Radue et al.7 applied structural 20 of them with clinical information. This report is image evaluation using normalization of atrophy to important for 3 reasons: (1) it confirms the sleep determine the effect of teriflunomide on brain vol- disorder as the core manifestation of the disease, ume loss (BVL) in patients with relapsing forms of (2) it describes previously unreported symptoms of MS enrolled in the phase 3 TEMSO study. Data CNS or peripheral nervous system hyperexcitability from 969 MRIs were examined, including 808 pa- (hyperekplexia, cramps, fasciculations, and stiffness) tients with baseline and 1-year MRI and 709 patients in 35% of the patients, and (3) it suggests that the with baseline and 2-year MRI treated either with disease may be more responsive to immunotherapy placebo or teriflunomide. The findings show that than previously thought. This is supported by the BVL was significantly lower in patients treated with symptom improvement observed in the case report teriflunomide (14 mg) vs placebo at year 1 and year 2, of Bonello et al.,5 also included in this issue of respectively. Teriflunomide at 7-mg dosing was also Neurology: Neuroimmunology & Neuroinflammation, associated with significant reduction in BVL vs pla- who used several immunotherapies (steroids, IV cebo at year 2. The authors conclude that these

From the ICREA-IDIBAPS, Hospital Clínic, University of Barcelona, Spain; and Department of Neurology, University of Pennsylvania. Funding information and disclosures are provided at the end of the editorial. Go to Neurology.org/nn for full disclosure forms. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 findings, coupled with published work in animal STUDY FUNDING models, suggest a potential neuroprotective activity No targeted funding reported. for teriflunomide and support a link between BVL and disability worsening. DISCLOSURE Derfuss et al.8 determined the time course of a4- J. Dalmau is the editor of Neurology: Neuroimmunology & Neuroinflam- mation; is on the editorial board for Neurology® and UpToDate; holds integrin receptor desaturation and disease activity in patents for and receives royalties from Ma2 autoantibody test, NMDA patients with relapsing-remitting MS who discontin- receptor autoantibody test, GABA(B) receptor autoantibody test, GABA ued natalizumab in the course of the TOFINGO trial, (A) receptor autoantibody test, DPPX autoantibody test, and IgLON5 a 32-week, patient and rater-masked multicenter, autoantibody test; and receives research support from Euroimmun, NIH, Fundació CELLEX, and Instituto Carlos III (CIBERER, and Fondo de parallel-group study. The study included 142 patients Investigaciones Sanitarias). Go to Neurology.org/nn for full disclosure who underwent measurement of a4-integrin receptor forms. occupancy (RO) at baseline and regular time periods until 24 weeks. Patients were randomized (1:1:1) to 8-, REFERENCES 12-, and 16-week washout groups. The findings show 1. Gaig C, Graus F, Compta Y, et al. Clinical manifesta- that a faster decline in natalizumab RO, longer wash- tions of the anti-IgLON5 disease. Neurology 2017;88: 1736–1743. out period, and higher T2 lesion volume at baseline 2. Graus F, Santamaria J. Understanding anti-IgLON5 dis- defines a population with an increased risk of return of ease. Neurol Neuroimmunol Neuroinflamm 2017;4:e393. inflammatory disease activity. According to the au- doi: 10.1212/NXI.0000000000000393. thors, these findings, along with the main outcomes 3. Gelpi E, Hoftberger R, Graus F, et al. Neuropathological of the TOFINGO study, support initiation of fingo- criteria of anti-IgLON5-related tauopathy. Acta Neuropa- – limod within 8 weeks of natalizumab discontinuation. thol 2016;132:531 543. 4. Honorat JA, Komorowski L, Josephs KA et al. IgLON5 Varicella zoster virus (VZV) vasculopathy is caused antibody: neurological accompaniments and outcomes in by productive virus infection of intracerebral arteries 20 patients. Neurol Neuroimmunol Neuroinflamm 2017; leading to stroke or aneurysm. Recent studies have 4:e385. doi: 10.1212/NXI.0000000000000385. shown that VZV can also affect extracranial arteries, with 5. Bonello M, Jacob A, Ellus A, et al. IgLON 5 disease respon- VZV antigen detected in 70% of patients with giant-cell sive to immunotherapy. Neurol Neuroimmunol Neuroin- arteritis.9 However, while the presence of VZV along flamm 2017;4:e383. doi: 10.1212/NXI.0000000000000383. 6. Solomon AJ, Watts R, Dewey BE, Reich DS. MRI eval- with inflammation has been demonstrated in cerebral uation of thalamic volume differentiates MS from com- and temporal arteries in patients with these disorders, mon mimics. Neurol Neuroimmunol Neuroinflamm VZV as the direct cause of the arterial inflammation 2017;4:e387. doi: 10.1212/NXI.0000000000000387. hasnotbeenshown.Toaddressthisquestion,Jones 7. RadueE-W,SprengerT,GaetanoL,etal.Teriflunomideslows et al. investigated whether VZV infection induced proin- BVL in relapsing MS: a reanalysis of the TEMSO MRI data set flammatory cytokines that result in arterial inflammation using SIENA. Neurol Neuroimmunol Neuroinflamm 2017;4: using several human brain vascular cell lines. Compared e390. doi: 10.1212/NXI.0000000000000390. 8. Derfuss TJ, Kovarik JM, Kappos L. a4-integrin receptor with mock infection, VZV infection led to significantly desaturation and disease activity return after natalizumab increased levels of interleukin 8 (IL8), IL6, and other cessation. Neurol Neuroimmunol Neuroinflamm 2017;4: cytokines in all or several of the cell lines.10 The authors e388. doi: 10.1212/NXI.0000000000000388. indicate that the VZV-mediated increase in IL8 and IL6 9. Gilden D, White T, Khmeleva N, Boyer PJ, Nagel MA. is consistent with that seen in the CSF of patients with VZV in biopsy-positive and -negative giant cell arteritis: 1 intracerebral VZV vasculopathy, and the arteries and analysis of 100 temporal arteries. Neurol Neuroimmu- nol Neuroinflamm 2016;3:e216. doi: 10.1212/NXI. blood of patients with giant-cell arteritis. Together with 0000000000000216. a previous study showing that VZV-infected arteries 10. Jones D, Neff CP, Plamer BE, Stenmark K, Nagel MA. downregulate PD-L1 to promote persistent inflamma- Varicella zoster virus–infected cerebrovascular cells pro- tion,11 the findings demonstrate that VZV infection is duce a proinflammatory environment. Neurol Neuro- sufficient to promote a proinflammatory environment immunol Neuroinflamm 2017;4:e382. doi: 10.1212/NXI. that may potentially lead to a persistent vasculitis. 0000000000000382. 11. Jones D, Blackmon A, Neff CP, et al. Varicella-zoster virus In addition to these studies, the September issue of downregulates programmed death ligand 1 and major his- Neurology: Neuroimmunology & Neuroinflammation tocompatibility complex class I in human brain vascular contains other interesting articles that I hope will adventitial fibroblasts, perineurial cells, and lung fibro- catch your attention. blasts. J Virol 2016;90:10527–10534.

2 Neurology: Neuroimmunology & Neuroinflammation EDITORIAL Understanding anti-IgLON5 disease

Francesc Graus, MD, The unambiguous characterization of antibodies of PSP, and cognitive deterioration sometimes asso- PhD against neural antigens has been critical to understand ciated with chorea. Independently of the form of pre- Joan Santamaría, MD, the interplay between the immune system and the sentation, most patients described a sleep disorder PhD brain, allowing in some instances the discovery of when directly questioned about symptoms of sleep clinical entities that had been previously unrecog- apnea and other alterations of sleep or when they were nized or lumped with other syndromes under generic studied with video-polysomnography (V-PSG).5 Correspondence to names.1 The best example is anti-NMDA receptor One potential bias of these initial studies is that Dr. Graus: [email protected] encephalitis, a relatively common disease that before they emphasized the sleep dysfunction, and therefore, the discovery of NMDA receptor antibodies was patients could have been selected for IgLON5 anti- Neurol Neuroimmunol unrecognized or defined with nonspecific terms, body screening. Recently, the article by Honorat Neuroinflamm such as non–herpetic encephalitis, or confused with et al.6 published in this issue of Neurology® Neuroim- 2017;4:e393; doi: 10.1212/ 2 NXI.0000000000000393 Hashimoto encephalopathy or encephalitis lethargica. munology and Neuroinflammation contributes to clarify A similar scenario applies to anti-IgLON5 disease this issue. These authors re-examined 367 archived in which the identification of antibodies against serum or CSF samples that showed a pattern of mice IgLON5, a neural cell adhesion molecule of brain immunostaining similar to that reported for unknown function, has led to the characterization IgLON5.3 Most of the samples had been sent for anti- of a new disorder.3 Before the discovery of IgLON5 body testing before the discovery of IgLON5 antibod- antibodies, the disease of these patients was defined ies. Using a commercial cell-based assay, they with descriptive terms or labeled as “atypical early eventually found that 26 of the samples had IgLON5 progressive supranuclear palsy (PSP)” or “brainstem antibodies. For 20 of these patients, clinical informa- tauopathy”(for review, see reference 4), indicating tion was available, including 15 who had detailed that it could not be classified among well-defined Mayo Clinic records. The findings confirmed that CNS disorders. However, the comprehensive evalua- brainstem and sleep disorders are the predominant tion of the initial patients with IgLON5 antibodies manifestations of anti-IgLON5 disease. Twelve of 15 disclosed a common clinical core of symptoms that patients developed sleep dysfunction, mainly sleep included a complex sleep disorder with abnormal be- apnea. The frequency of abnormal movement and be- haviors and vocalizations during non-REM and REM haviors was lower, but most of the patients were not sleep along with sleep-disordered breathing, gait assessed by V-PSG, which in our experience is critical instability, more typical of disequilibrium than cere- for full assessment of this disease. In addition, 14/20 bellar ataxia, and bulbar symptoms such as stridor, (70%) patients showed gait instability and 12 dyspha- dysphagia, or central hypoventilation.3 These symp- gia or stridor. Unlike previous series, 7 (35%) patients toms were found to evolve over months to years, and had symptoms of central or peripheral nervous system unlike other CNS disorders associated with antibod- hyperexcitability, including hyperekplexia, fascicula- ies against neural surface antigens, patients with tions, cramps, or stiffness. Therefore, this study con- IgLON5 disease did not respond to common immu- firms that the core symptoms of anti-IgLON5 disease notherapies. With the identification of more patients, include the sleep disorder along with gait instability 4 clinical presentations were characterized according and bulbar dysfunction and draws attention to inves- to predominant symptoms: prominent sleep disorder, tigate for this disease in patients with symptoms that progressive bulbar dysfunction, gait instability with suggest, but are unusual for, stiff-person syndrome abnormal eye movements reminiscent, but not typical spectrum disorder.7

From the Neuroimmunology Program (F.G.), August Pi Sunyer Biomedical Research Institute (IDIBAPS); and Service of Neurology (F.G., J.S.) and Multidisciplinary Sleep Disorders Unit (J.S.), Hospital Clínic, Barcelona, Spain. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by Neurology: Neuroimmunology & Neuroinflammation. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 An intriguing aspect of anti-IgLON5 disease is the neurologic disorders has driven to the development concurrence of features suggesting an autoimmune of multiple neurologic subspecialties, “Autoimmune disorder (antibodies against a neuronal surface anti- Neurology” one of the youngest. Specialization is gen and a robust association with the HLA- inevitable, but it should not lead to fragmentation DRB1*10:01 genotype) and pathologic evidence of of knowledge or tight boundaries; anti-IgLON5 dis- a novel neuronal 3R 1 4R tauopathy mainly involv- ease is a perfect example that cross-talk between sub- ing the brainstem and hypothalamus.4 This puzzling specialties is critical to address complex neurologic scenario challenges current concepts about the path- diseases and, above all, that a solid neurologic back- ophysiology of autoimmune and neurodegenerative ground is the key to unlock puzzling clinical disorders. One possibility is that the neuronal- scenarios. specific tau accumulation could be antibody medi- ated, leading to neuronal dysfunction and ultimately AUTHOR CONTRIBUTIONS neurodegeneration.8 In this scenario, early diagnosis Francesc Graus: drafting/revising the manuscript, study concept or design, and study supervision. Joan Santamaría: drafting/revising the and treatment would be crucial to prevent irreversible manuscript. neuronal damage. The patients described in initial studies did not improve with immunotherapy, but STUDY FUNDING there was a substantial delay (median, 2 years) No targeted funding reported. between symptom onset and diagnosis.5 The retro- See Article spective study of Honorat et al.6 and the case report of DISCLOSURE Bonello et al.9 in this issue of Neurology Neuroimmu- F. Graus is on the editorial board for Lancet Neurology, receives royalties, nology and Neuroinflammation also suggest that and holds a patent for IgLON5 as a diagnostic test. J. Santamaría See Article received research support from Fundacio La Marato TV3. Go to anti-IgLON5 disease may indeed respond to immu- Neurology.org/nn for full disclosure forms. notherapy. Seven of 10 patients in the former study improved after treatment with steroids alone or in REFERENCES combination with other immunotherapies.6 These 1. Dalmau J, Geis C, Graus F. Autoantibodies to synaptic findings, however, must be taken with caution, given receptors and neuronal cell surface proteins in autoim- the retrospective accrual of information and lack of mune diseases of the central nervous system. Physiol Rev 2017;97:839–887. follow-up in half of the patients. In addition, poten- 2. Tan A, Shuey N, Bladin C. A modern perspective on the tial confounding factors such as the use of continuous differential diagnosis between encephalitis lethargica or positive airway pressure therapy to improve the sleep anti-NMDA-receptor encephalitis. J Clin Neurosci 2010; apnea must be taken into consideration at the time of 17:1204–1206. evaluation of response to treatment. The patient 3. Sabater L, Gaig C, Gelpi E, et al. A novel non-rapid-eye described by Bonello et al.9 also improved with treat- movement and rapid-eye-movement parasomnia with ment that started 1 year after disease onset. This sleep breathing disorder associated with antibodies to IgLON5: a case series, characterisation of the antigen, patient developed the typical syndrome of anti- and post-mortem study. Lancet Neurol 2014;13:575–586. IgLON5 disease, but was younger (45 years) than 4. Gelpi E, Höftberger R, Graus F, et al. Neuropathological most patients (median age 62 years) with this disor- criteria of anti-IgLON5-related tauopathy. Acta Neuropa- der. If younger patients are more likely to improve is thol 2016;132:531–543. currently unclear because the median age of the 7 5. Gaig C, Graus F, Compta Y, et al. Clinical manifestations of – patients who improved in the series of Honorat the anti-IgLON5 disease. Neurology 2017;88:1736 1743. 6. Honorat JA, Komorowski L, Josephs KA, et al. IgLON5 et al.6 was similar to that of the whole series. antibody: neurological accompaniments and outcomes in Future series and single case reports of patients 20 patients. Neurol Neuroimmunol Neuroinflamm 2017; with IgLON5 disease will be important to define 4:e385. doi: 10.1212/NXI.0000000000000385. the effects of immunotherapy, identify less com- 7. Martinez-Hernandez E, Ariño H, McKeon A, et al. Clin- mon clinical presentations, and better understand ical and Immunologic investigations in patients with whether IgLON5 antibodies occur as an epiphe- stiff-person spectrum disorder. JAMA Neurol 2016;73: 714–720. nomenon in some neurologic disorders. Prelimi- 8. Leshchyns’ka I, Sytnyk V. Reciprocal interactions between nary studies suggest that IgLON5 antibodies cell adhesion molecules of the immunoglobulin superfam- cause an irreversible antibody-mediated internaliza- ily and the cytoskeleton in neurons. Front Cell Dev Biol tion of surface IgLON5 in cultures of hippocampal 2016;4:9. neurons.10 The natural next step is to determine 9. Bonello M, Jacob A, Ellul MA, et al. IgLON5 disease respon- whether this antibody effect leads to a disruption sive to immunotherapy. Neurol Neuroimmunol Neuroin- flamm 2017;4:e383. doi: 10.1212/NXI.0000000000000383. of the internal cytoskeletal network and ultimately 10. Sabater L, Planagumà J, Dalmau J, Graus F. Cellular in- neuronal tau accumulation. vestigations with human antibodies associated with the The field of neurologic sciences is wide and ex- anti-IgLON5 syndrome. J Neuroinflammation 2016; panding, and the management of patients with 13:226.

2 Neurology: Neuroimmunology & Neuroinflammation IgLON5 antibody Neurological accompaniments and outcomes in 20 patients

Josephe A. Honorat, MD, ABSTRACT PhD Objective: To describe the phenotypes, treatment response, and outcome of IgLON5 Lars Komorowski, PhD autoimmunity. Keith A. Josephs, MD Methods: Archived serum and CSF specimens from 367 patients known to harbor unclassified Kai Fechner, MEng antibodies which stained neural synapses diffusely (mimicking amphiphysin-IgG) were reeval- Erik K. St Louis, MD, MS uated by indirect immunofluorescence assay (IFA) using a composite of mouse tissues and re- Shannon R. Hinson, PhD combinant IgLON5-transfected cell-based assay (CBA, Euroimmun). Sabine Lederer Neeraj Kumar, MD Results: Available specimens (serum, 25; CSF, 9) from 26/367 patients (7%) had identical IFA Avi Gadoth, MD appearance and robust IgLON5 CBA positivity. Clinical information was available for 20/26 pa- Vanda A. Lennon, MD, tients; 13 were women. Median disease-onset age was 62 years (range, 46–75 years). Most PhD patients had insidious onset and progression of neurological symptoms affecting movement and Sean J. Pittock, MD sleep predominantly. Sleep disorders were sleep-disordered breathing (11) and parasomnias (3). Andrew McKeon, MD Brainstem disorders were gait instability (14), dysphagia (10), abnormal eye movements (7), respiratory dysfunction (6), ataxia (5), craniocervical dystonia (3), and dysarthria (3). Findings compatible with hyperexcitability included myoclonus (3), cramps (3), fasciculations (2), and exag- Correspondence to gerated startle (2). Neuropsychiatric disorders included cognitive dysfunction (6), psychiatric Dr. McKeon: symptoms (5), and seizures (1). Dysautonomia, in 9, affected bladder function (7), gastrointestinal [email protected] motility (3), thermoregulation (3), and orthostatic tolerance (1). Just 2 patients had coexisting autoimmune disease. Brain MRI findings were nonspecific and CSF was noninflammatory in all tested. Seven of 9 immunotherapy-treated patients improved: 6 of those 7 were stable at last follow-up. Three untreated patients died. Each IgLON5-IgG subclass (1–4) was readily detectable in $80% of specimens using CBA. Conclusions: IgLON5-IgG is diagnostic of a potentially treatable neurological disorder, where autoimmune clues are otherwise lacking. Neurol Neuroimmunol Neuroinflamm 2017;4:e385; doi: 10.1212/NXI.0000000000000385

GLOSSARY AHI 5 Apnea-Hypopnea Index; ANNA 5 antineuronal nuclear antibody; CBA 5 cell-based assay; GAD65 5 glutamic acid decarboxylase 65; MSA 5 multiple system atrophy; OSA 5 obstructive sleep apnea; PSP 5 progressive supranuclear palsy.

Since 2004, several autoimmune CNS disorders have been described, each characterized serolog- ically by an antibody targeting the extracellular domain of a neural autoantigen.1 Pertinent exam- ples include aquaporin-4, NMDA receptor (NR1 subunit), and glycine receptor.2–4 Subacute onset and rapid symptom progression, and responses to immune therapies are typical for these disorders, although not universal.5 Autoimmunity targeting the extracellular domain of an immunoglobulin- like cell adhesion molecule, IgLON5, is a reported exception to this general rule.6–8 IgLON5 is an adhesion molecule widely expressed in the CNS, but whose precise function is Supplemental data poorly understood. IgLON5 autoimmunity has been characterized as a progressive CNS at Neurology.org/nn From the Department of Laboratory Medicine and Pathology (J.A.H., S.R.H., V.A.L., S.J.P., A.M.), Department of Neurology (K.A.J., E.K.S.L., N.K., A.G., V.A.L., S.J.P., A.M.), Department of Medicine (E.K.S.L.), Department of Immunology (V.A.L.), and Center for Sleep Medicine (E.K. S.L.), College of Medicine, Mayo Clinic, Rochester, MN; and Institute of Experimental Immunology (L.K., K.F., S.L.), Euroimmun AG, Lubeck, Germany. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 disorder of insidious onset with prominent based assay (CBA) on human embryonic kidney 293 cells that sleep and movement abnormalities. Progres- were transfected with the IgLON5 complementary DNA, fixed with 1% formalin and stored at 4°C (Euroimmun AG, Lubeck, sion to death due to respiratory failure is com- Germany). The sera (1:10 dilution) or CSF (neat) was incubated mon.6 Reported neuropathologic findings of with the transfected cells. The cells were then washed and exposed “tauopathy,” which unifies a number of disor- to fluorescein isothiocyanate–conjugated goat antihuman IgG ders, such as progressive supranuclear palsy (Southern Biotech, Birmingham, AL). For IgG subclass testing, antihuman secondary antibodies specific for IgG1 (Thermo (PSP), and corticobasal degeneration, have Fisher, Waltham, MA), IgG2 (Southern Biotech), IgG3 (In- also reinforced the concept of IgLON5 auto- vitrogen, Carlsbad, CA and The Binding Site, Birmingham, AL), immunity having a poor prognosis.9 Charac- and IgG4 (Southern Biotech) were used. teristic neuropathologic findings include a lack Patients. We evaluated 367 archived specimens (serum, 334; of inflammatory infiltrates, neuronal loss, glio- CSF, 33) from patients referred to the Mayo Clinic Neuroimmu- nology Laboratory for evaluation of neurological autoimmunity sis, and neuronal accumulation of hyperphos- (1997–2016), for whom tissue-based immunofluorescence assay phorylated tau protein (both 3-repeat and (IFA) screening had demonstrated a pattern of diffuse neural 4-repeat [3R 1 4R] isoforms) found predom- synaptic (neuropil) staining resembling (but not meeting criteria inantly in the hypothalamus and the tegmen- for) amphiphysin-IgG (which is similar to that reported for Ig- LON5-IgG).6 Specimens were retested by IFA and by CBA for 9 tal brainstem nuclei. IgLON5-IgG. In this article, we report our Mayo Clinic Sera and CSF from 96 patients diagnosed at Mayo Clinic experience of IgLON5 autoimmunity in 20 pa- with the following disorders that might mimic IgLON5 auto- immunity were also evaluated by IFA and CBA for IgLON5- tients, including neurological improvements, in IgG: PSP, 50; multiple system atrophy (MSA), 40; progressive response to immune therapies. age-related tauopathy, 3; and glutamic acid decarboxylase 65 (GAD65) kDa isoform neurological autoimmunity mimicking METHODS Standard protocol approvals, registrations, PSP, 3. and patient consents. The Mayo Clinic Institutional Review Of the 26 patients determined to be Board approved this study (no. 08-006647). Clinical evaluation. IgLON5-IgG positive by IFA and CBA, clinical information Serological testing. Specimens (serum and CSF) were evalu- was available in 20. Fifteen had a Mayo Clinic medical record that ated by indirect immunofluorescence assay (IFA) on a composite was reviewed extensively; comprehensive neurological evaluations substrate of mouse hippocampus, cerebral cortex, cerebellum, were available for all (including history, physical examination basal ganglia, thalamus, kidney, and stomach. Four-micrometer findings, laboratory results, imaging data, and electrophysiologic frozen cryosections were fixed, blocked, and incubated with testing). Patients from elsewhere (5) had data limited primarily to patient serum or CSF for 40 minutes and then with appropriate the neurological phenotype available and were not included in secondary antibodies. IgLON5 specificity was confirmed by cell- our analysis of sleep data. Among Mayo Clinic patients, poly- somnography was available for review in 3 patients permitting detailed review of video-PSG findings, and portable overnight oximetry was available for 3 patients (including 1 patient who did Figure 1 IgLON5 antibody tissue-based indirect immunofluorescence assay not have polysomnography). Sleep diagnoses or reports of sleep symptoms were available only from medical records for an additional 10 patients, made prior to Mayo Clinic evaluation.

RESULTS Serological results. All 26 patients had the IgLON5-IgG characteristic pattern of staining by indirect IFA, and positivity was confirmed for all specimens by IgLON5 CBA. In the laboratory, 19 patients were identified retrospectively, 7 pro- spectively after initiation of the study (March 2016– January 2017), approximately 1 per month. Two of those 7 were evaluated in person by AM. The syn- aptic pattern of immunoreactivity was more intense in the cerebellum (granular layer more than molecular layer), midbrain, and thalamus than in the hippo- campus and cerebral cortex (figure 1). The renal glo- meruli and the smooth muscle of the stomach (but not myenteric nervous tissue) were also reactive with Immunoreactivity of patient serum IgLON5-IgG is more prominent in the cerebellum (A) than patient IgG in all cases. Unlike smooth muscle anti- in the hippocampus (B). IgLON5-IgG immunoreactivity is present also in intestinal smooth body, preabsorption of serum with bovine liver powder muscle (C) and renal glomeruli (D). GL 5 granular layer of cerebellum; GM 5 renal glomeruli; H 5 hippocampus; M 5 intestinal mucosa; ML 5 molecular layer of cerebellum; SM 5 intes- did not abrogate the smooth muscle staining by Ig- tinal smooth muscle layer; T 5 thalamus. Scale bar: 0.5 mm. LON5-IgG. The median IgLON5-IgG antibody value

2 Neurology: Neuroimmunology & Neuroinflammation in serum was 1:3,840 (range 1:480–1:15,360; normal available for testing in 19 and CSF in 8; all specimens value #120). All 4 IgG subclasses were detected using were positive for IgLON5-IgG by both IFA and CBA in most patients tested (20/24): IgG1, 23/24; CBA. Symptoms evolved in an insidious and pro- IgG2, 20/24; IgG3, 23/24; and IgG4, 22/24. Fluo- gressive manner in 15 patients (over several months to rescence was more intense for IgG1 and IgG4 than years) and subacutely in 5 patients (less than 1 IgG2 and or IgG3 (figure 2). Just a single patient had 1 month). The median duration of the disease from the IgLON5-IgG subclass detected (IgG4, patient 15). onset of symptoms to last follow-up was 30 months Two patients had coexisting antibodies detected in (range, 2–156 months). serum (leucine-rich glioma-inactivated 1 antibody, 1 and Clinical details are summarized in table 1. The GAD65 antibody, 1). None of the 96 control patients predominant neurological manifestations of IgLON5 (degenerative disease or degenerative disease mimics) autoimmunity were brainstem disorders (dysphagia, were IgLON5 antibody positive by IFA or CBA. dysarthria, eye movement disorders, and gait instabil- ity), sleep disorders (including sleep-disordered Summary of demographic and clinical findings. Thirteen breathing and parasomnias), neuropsychiatric disor- of the 26 patients were women (50%); all were resid- ders (cognitive impairment and psychiatric symptoms), ing in the United States. All 14 patients where infor- dysautonomia, and hyperexcitability (myoclonus, mation regarding race was available were white. The cramps, and exaggerated startle). median age of symptom onset was 62 years (range, 46–75 years). Clinical information was available for Clinical course. Sleep disorders. Sleep disorders or sleep 20 patients. For those 20 patients, serum was symptoms were reported in 12 of 15 patients where that history was available (75%, table 1), 5 of whom Figure 2 IgLON5 antibody cell-based assay and IgG subtypes were known to have had sleep symptoms first. Of note, all 3 patients evaluated prospectively after the IgLON5 antibody serologic finding presented with sleep disorders. Obstructive sleep apnea (OSA) was diagnosed in 11 patients (although only 2 had this diagnosis confirmed by polysomnography in the course of their evaluations at Mayo Clinic, neither of these patients had audible stridor throughout sleep nor during wakefulness, and there was no report of recorded sleep or waking stridor nor history consis- tent with stridor provided by patients or their families in the remaining patients). Five patients had sleep apnea prior to the onset of neurological symptoms without temporal association to other neurological symptoms. A further 6 patients had the diagnosis of sleep apnea performed at the time of initial neurolog- ical evaluation or subsequently. Of 4 patients with sleep apnea discovered prospectively with known apnea severity, 3 were severe (Apnea-Hypopnea Index [AHI] 60, 47, and 38/hr), and 1 was moderate (AHI 15.4/hr), and all patients with severe OSA were trea- ted successfully with nasal continuous positive airway pressure therapy, whereas the patient with moderate OSA was advised to obtain sleep medicine follow-up following a diagnostic study but was lost to further follow-up. Three patients had parasomnias diagnosed fol- lowing video-PSG review at our institution (by EKS). One was reported to have dream enactment behavior but had no recorded REM during poly- somnogram. The second was reported to have dream enactment and had REM sleep without atonia. The IgLON5-IgG subclasses were characterized in patient serum. Goat antihuman IgGs (reactive third had an NREM parasomnia disorder of arousal with all IgG [A] or 1 IgG subclass [C–F]) were used as secondary antibodies. IgG reactivity was more prominent for IgG1 (C) and IgG4 (F) than for IgG2 (D) or IgG3 (E). Patient IgG was (confusional arousals from N2 sleep during nonreactive with mock-transfected cells (B). Scale bar: 0.5 mm. polysomnography).

Neurology: Neuroimmunology & Neuroinflammation 3 4

Table 1 Demographic and clinical data for 20 IgLON5 autoimmune patients

Peripheral Patient no./ Gait Movement Abnormal eye Neuropsychiatric nervous sex/age Initial findings instability disorders movements Bulbar symptoms Sleep disorders symptoms Dysautonomia system

1/F/69a Headache; jaw dystonia Yes Parkinsonism, myoclonus; Supranuclear gaze No No Depression; memory No No myorhythmia (mouth, tongue); palsy (up gaze impairment cervicofacial dystonia; tremor predominant)

2/M/75a Myoclonus, diplopia; Yes Myoclonus; parkinsonism Hypometric No OSA; NFLE (screaming; Memory impairment Urinary, bowel No cognitive impairment, saccades grabbing) incontinence; anhidrosis erlg:Nuomuooy&Neuroinflammation & Neuroimmunology Neurology: sleep apnea

3/F/62a Gait instability, Yes Spastic gait; choreoathetosis; Nystagmus Dysphagia; aspiration OSA No Urinary incontinence No extremity weakness cervical dystonia

4/F/67a Dysphagia No Adventitious movements (tongue); No Dysphagia OSA Hallucinations; Delirium; No No jaw dystonia impaired memory

5/F/72a Gait instability; dizziness Yes Parkinsonism; ataxia Nystagmus; Dysphagia No Subtle executive No No supranuclear gaze dysfunction palsy (up gaze)

6/F/70a Gait instability; Hx of Yes Tremor; parkinsonism; myoclonus No Respiratory failure, OSA; parasomnia Hallucinations; delirium; Urinary incontinence No sleep apnea stridor, laryngospasms (moaning, talking, kicking; depression; memory RBD) impairment

7/M/63a Diplopia, headache No No No Orthostatic hypotension; OSA Anxiety Urinary hesitancy No dysphagia

8/M/50a Gait instability Yes Ataxia No No No No Anhidrosis Cramps

9/F/64 Gait instability Yes Ataxia Nystagmus Dysphagia Unknown No No PN

10/M/61a Dysphagia Yes Ataxia No Dysphagia; aspiration OSA Confusion; hallucinations Urinary frequency PN

11/F/66a Gait instability; diplopia; Yes Parkinsonism Horizontal gaze Dysphagia; respiratory OSA No No PN Hx of sleep apnea paresis failure

12/M/59a Cramps, fasciculations; No No Ptosis No OSA No No Cramps Hx of sleep apnea

13/M/61a Weakness, paresthesias; Yes No No No OSA No No PN Hx of sleep apnea

14/F/52a,b Muscle spasm and Yes Stiff-person syndrome No No OSA, RBD No No No stiffness

15/F/61 Laryngospasm No No Supranuclear gaze Laryngospasms Unknown No Urinary retention Fasciculations palsy

16/M/46a Fasciculations, diplopia No No No Aspiration; dysphagia OSA No Intestinal dysmotility; Cramps and anhidrosis fasciculations

17/F/59 Paresthesia Yes No No No Unknown No No No

18/M/54 Dysphagia, diplopia Yes Ataxia No Dysphagia Unknown No No No

19/M/72 Confusion No No No No Unknown No No No

20/F/59a,b Dysphagia, cognitive Yes Chorea No Dysphagia Non-REM parasomnias Memory impairment, Urinary incontinence; No impairment (confusional arousals) disinhibition constipation

Abbreviations: Hx 5 history; NFLE 5 nocturnal frontal lobe epilepsy; OSA 5 obstructive sleep apnea; PN 5 peripheral neuropathy; RBD 5 REM sleep behavior disorder. a Evaluated clinically at Mayo Clinic. b Evaluated prospectively by AM. One patient had stereotyped sleep-related complex Other diagnoses. One patient had adenocarcinoma motor behaviors diagnostic of nocturnal frontal lobe of the breast detected prospectively. An additional 2 epilepsy (documented electrographic seizures corre- patients had a history of cancer (prostate adenocarci- sponding to the clinical phenomena [shouting, fol- noma, 1 [detected 5 years prior to the onset of neuro- lowed by right arm jerking, and then stiffening] in logical symptoms] and remote history of non-Hodgkin the epilepsy monitoring unit). lymphoma, 1). Two patients reported a history of Neurological disorders. Fourteen of 20 (70%) pa- autoimmune disease (one each of Hashimoto thy- tients had gait instability accompanied by ataxia, 5; roiditis and recurrent iritis). chorea, 2; parkinsonism, 5 (akinetic, rigid phenotype CSF studies. CSF testing was largely unremarkable. in all); or limb and axial stiffness, 1. Falls were fre- Five patients had a mild elevation of protein concen- quent in 5 patients, 3 of whom were diagnosed with tration (median value, 57 mg/dL: range, 49–106 mg/ a neurodegenerative disorder by movement disorders dL [normal, 0–35 mg/dL]). None had elevations in subspecialists at our institution (PSP, 1; MSA, 1; or white blood cells (more than 5/mL), IgG indices synucleinopathy not otherwise specified, 1). The (.0.85), IgG synthesis rates (.12 mg/24 hours), patient diagnosed with PSP presented with early falls or CSF-exclusive oligoclonal band numbers ($4). and was found to have axial more than appendicular rigidity and symmetric bradykinesia but did not have Imaging and electrophysiology. Brain MRIs were also vertical gaze palsy. None of the other patients had iso- largely unremarkable: normal in 10 patients and dem- lated downgaze palsy or both up and downgaze paral- onstrated one or more nonspecific abnormalities in yses typical of the PSP phenotype (Richardson 10: leukoariotic changes, 6; mild-to-moderate cere- syndrome). bral atrophy, 6; and cerebellar atrophy, 2 (both had Other brainstem symptoms comprised dysphagia, ataxia). No parenchymal spinal cord abnormalities 10; respiratory dysfunction, 6 (3 of whom developed were noted in any of 13 patients tested. EEG was laryngospasm or respiratory failure requiring tracheos- abnormal in 1 patient with frontal lobe seizures. tomy and 1 also had stridor); pulmonary aspiration, 3; EMG revealed abnormal findings in 3 patients: axo- diplopia, 5; and dysarthria, 3. Hyperkinetic movement nal sensorimotor peripheral neuropathy, 1; cramps disorders, present in 7, were myoclonus, 3; craniocer- and fasciculations, 1; and mononeuritis multiplex, 1. vical dystonia, 3; tremor, 2; myorhythmia, 1; chorea, EMG study using surface electrodes in the Movement 1; and choreoathetosis, 1. Patient 1 had an akinetic Disorders Electrophysiology Laboratory study dem- rigid syndrome with supranuclear gaze palsy and oro- onstrated cortical myoclonus in 1 patient and exag- lingual myorhythmia. Whipple disease was considered gerated exteroceptive response in another (patient 14 but excluded. Patient 20 (the video at Neurology.org/ with stiff-person syndrome). nn) was initially thought to have a functional disorder Treatment and outcomes. Ten patients received one or because of behavioral abnormalities in addition to more immunotherapies and had follow-up available a movement disorder but was later recognized to have (table 2). Improvements were encountered in 5 of 7 a frontal-subcortical syndrome with chorea. patients who received steroids, all 3 who received Neuropsychiatric disorders affecting 8 patients plasma exchange, both who received IVIg, both who (40%) included memory loss, 5; delirium, 4; halluci- received azathioprine, 1 of 2 who received myco- nations, 3; depression, 2; and executive dysfunction, phenolate mofetil, and 1 who received rituximab. The 1. One patient had frontal lobe seizures. improvements resulted in sustained near full recovery, Findings consistent with central or peripheral ner- 2; sustained partial improvements, 5; and initial vous system hyperexcitability were noted in 7 patients improvement with subsequent worsening, 1. Two (35%). Patient 14 had hyperekplexia (paroxysmal patients had no response to immunotherapy, one of whole-body spasms in response to startle) and was diag- whom eventually died (table 2). Four patients in total nosed with a stiff-person syndrome spectrum disorder. were reported to have died, 3 had not received Seven patients had upper motor neuron signs: hyperre- immunotherapy. The median modified Rankin score flexia, 8, extensor plantar responses, 3; and spasticity, 1. at last follow-up was 2.5 (range, 1–6); median dura- Nine patients (45%) had autonomic dysfunction tion from evaluation to last follow-up was 15 months affecting bladder, 7; gastrointestinal motility (diar- (range, 0.5–129 months). rhea or constipation), 3; sweating (anhidrosis), 3; and orthostatic hypotension, 1. Four patients had evi- DISCUSSION IgLON5 antibody is a rare finding in dence of peripheral neuropathy. One had mononeur- our laboratory (1 per month), which evaluates itis multiplex and sural nerve biopsy-proven small approximately 150,000 specimens per annum. The vessel vasculitis, and the other 3 had EMG evidence clinical course of patients encountered was usually supporting peripheral neuropathy (axonal sensorimo- insidious in onset and progression. Neurological dis- tor, 2 and motor demyelinating, 1). orders tended to be multifocal, although consistent

Neurology: Neuroimmunology & Neuroinflammation 5 Table 2 Antibody findings, treatments, and outcomes for 20 patients

mRS score at Patient no./ IgLON5 Ab Duration of last sex/age detection Immunotherapy Response to immunotherapy follow-up, mo follow-up

1/F/69a S None NA 4 6 (death)

2/M/75a S, C Ster Resolution of seizures. Improved myoclonus 30 1 and cognitive dysfunctionc

3/F/62a S, C Ster, Aza Improved chorea and gait instability initially 88 4 maintained on Aza alone. Worsened on discontinuing. Improved again with Ster and Aza

4/F/67a S None NA 6 6 (death)

5/F/72a S None NA 14 3

6/F/70a S, C Steroids No response 19 6 (death)

7/M/63a S None NA 9 1

8/M/50a S None NA 0.5 2

9/F/64 S Ster, MM Improved ataxia and dysphagia. Maintained 48 4 on MM only

10/M/61a S None NA 129 3

11/F/66a C None NA 9 6 (death)

12/M/59a S None NA NA 1

13/M/61a S Ster, Cyc, Aza Improved weakness and paresthesias. 23 1 Maintained on Aza after Ster and Cyc

14/F/52a,b S Ster, IVIg, MM, Rit Improved muscle spasms and stiffness, 15 1 with IVIg and MM, but then relapsed on tapering IVIg. Response to Rit permitted remission and tapering of steroid

15/F/61 S, C NA NA NA 3

16/M/46a S IVIg, PLEX Improved cramps, fasciculations, and 15 2 intestinal dysmotility

17/F/59 SNANA NA3

18/M/54 S, C Ster No response NA 2

19/M/72 S, C PLEX Improved encephalopathy NA 1

20/F/59a,b S, C Ster, PLEX No improvement with steroids, improved 12 with PLEX (see video)

Abbreviations: Aza 5 azathioprine; C 5 CSF tested and positive for IgLON5-IgG; Cyc 5 cyclophosphamide; IVIg 5 IV immunoglobulin; MM 5 mycophenolate mofetil; mRS 5 modified Rankin scale; NA 5 not available; PLEX 5 plasma exchange; Rit 5 rituximab; S 5 Serum tested and positive for IgLON5-IgG; Ster 5 corticosteroids. a Evaluated clinically at Mayo Clinic. b Evaluated prospectively by AM. c Patient also given levetiracetam.

with prior reports, findings of brainstem encephalop- Generic clues to an autoimmune diagnosis (sub- athy (affecting primarily movement, gait, and bal- acute presentation, history of autoimmune disease, ance) and sleep were most common.6–8,10–13 Sleep an inflammatory CSF, or inflammatory-appearing disorders were frequent, but the clinical spectrum brain imaging) were infrequent, and thus, antibody seemed more heterogenous than previous reports of testing of serum or CSF for IgLON5-IgG was critical. patients presenting to sleep clinics. We encountered At this early stage, it would seem that serum and CSF both mild and severe sleep disorders, possibly due to have equal diagnostic sensitivity, although in general, our screening process (for antibody rather than for when screening for all these rare autoimmune clinical phenotype). It is possible that sleep disorders encephalitides in aggregate, it is recommended to test may have been overlooked in some patients (poly- both serum and CSF.14 somnography was absent in all but 3) or worsened We encountered diverse movement disorders, after the patients left Mayo Clinic (duration of including parkinsonism, ataxia, dystonia, chorea, follow-up was just 15 months). Notably, all 3 pa- and myoclonus.6,8,12 Patient 14, with stiff-person syn- tients evaluated prospectively by the coauthors had drome, did not have GAD65 antibody or glycine sleep disorders. receptor antibody detected, but responded to

6 Neurology: Neuroimmunology & Neuroinflammation immunotherapy nonetheless.15 Although not pheno- The discrepancies in outcome between our pa- typically classic, in some rare instances, IgLON5 tients and those previously reported may be explained autoimmunity may clinically resemble a tauopathy, by differences in case ascertainment. IgLON5 auto- such as PSP, or a synucleinopathy, such as MSA. The immunity was originally described among a cohort PSP-typical phenotypic vertical eye movement abnor- of patients identified by the clinical phenotype.6 By malities were absent in the case initially diagnosed as contrast, our patients were encountered on a serologic such in our series. Furthermore, the pathology of basis during a broad screen for neural antibodies using IgLON5 autoimmunity more resembles that encoun- tissue-based IFA. Sleep symptoms were often present, tered in primary age-related tauopathy (3R 1 4R but they were not always prominent, and respiratory isoforms) rather than PSP (4R tau isoforms only).16,17 failure occurred in a minority. Three patients had Patient 1 had myorhythmia and supranuclear gaze laryngospasm or respiratory failure requiring trache- palsy, resembling what has been reported for CNS ostomy, 2 of whom subsequently died. Laryngospasm Whipple disease.18 with sudden death is also reported among patients A paraneoplastic cause of IgLON5 autoimmunity with antineuronal nuclear antibody type 2 (ANNA- appears to be rare, although patients were not system- 2, Ri) autoimmunity. Other phenotypic similarities atically evaluated for the detected cancers (adenocar- with ANNA-2 autoimmunity are dystonia (including cinomas) during their evaluation at Mayo Clinic. At jaw dystonia), eye movement disorders, and parkin- this early stage, screening broadly, one time, for age, sonism.20 We plan to prospectively analyze a larger sex, and other risk factor appropriate neoplasms sample of patients with IgLON5 autoimmunity syn- among IgLON5-IgG–positive patients seems appro- drome in detail to more fully characterize the spec- priate, given the track record for most autoimmune trum of sleep and respiratory disorders in this patient neurological disorders having some degree of onco- population. logical significance.1 IgLONs are members of the immunoglobulin We observed a range of responses to treatments used superfamily of cell adhesion molecules abundantly ex- among IgLON5 autoimmune cases and a range of neu- pressed in neurons. The IgLON proteins contain 3 rological prognoses. The improvements we report are in immunoglobulin domains followed by a glycosylphos- contrast to the original report of this disorder and the phatidylinositol anchor protein.21 Metalloproteinase- largest series to date (22 cases) where robust immuno- dependent regulated shedding of IgLONs from the therapy responses were rare and death occurred in surface of mature cortical neurons has been reported 60%.6,13 Occasional individual cases of immunother- to regulate neurite outgrowth and synaptic plastic- apy responsiveness have been reported.11,12 However, ity.21 Our data support IgLON5 being abundant, sudden death, despite initial responses to immune treat- not just in the nervous system but also in the renal ments, has also been reported.11 glomerulus, which has afferent and efferent innerva- Similar to what has been reported previously, tion to regulate blood flow to the glomerular base- death was common,6 although in our patients, those ment membrane and a diverse repertoire of neural who had not received immune therapies faired worst. antigens.22–25 Like ANNA-3 autoimmunity, where Although the finding of tau deposition in the brain renal podocyte staining on tissue IFA is also recog- intuitively could lead one to conclude that IgLON5 nized, renal manifestations are not known to accom- autoimmunity is a secondary event in an otherwise pany IgLON5 autoimmunity.26 neurodegenerative disorder, other data point to In the appropriate clinical context (adults with IgLON5-IgG having pathogenic effects. In vitro unexplained sleep disorder, movement disorders, bul- treatment of hippocampal neurons with IgG1 (but bar symptoms, CNS hyperexcitability), IgLON5 not IgG4) from affected patients has been shown to autoimmunity should be considered. Generic clues cause internalization of IgLON5 clusters from the to an autoimmune diagnosis are lacking in most cases, neuronal cell surface.19 All IgLON5-IgG subclasses although responses to immune therapies may be (1–4) were detectable in most of our patients, favorable. although IgG1 and IgG4 were most intense by indirect immunofluorescent CBA. Our findings AUTHOR CONTRIBUTIONS essentially confirm that an initial report that dem- J.A.H.: data collection, analysis and interpretation, and drafting of man- uscript. L.K., K.F., and S.L.: supplied critical reagents and critical revi- onstrated, quantitatively, IglON5-IgG is predomi- sion of manuscript. K.A.J., E.K.S.L., S.R.H., N.K., A.G., V.A.L., and 19 nantly represented in IgG1 and IgG4 subclasses. S.J.P.: data collection and analysis and critical revision of manuscript. The lack of inflammatory findings in CSF in our A.M.: study concept and design, data collection, analysis, and interpreta- patients, and at autopsy in prior reports, may sup- tion, critical revision of manuscript, and study supervision. port IgG1-mediated IgLON5 internalization, ACKNOWLEDGMENT without complement activation, as a key patho- The authors thank the Mayo Clinic Center for Individualized Medicine physiologic mechanism.6,9 and the Department of Laboratory Medicine and Pathology, Mayo Clinic,

Neurology: Neuroimmunology & Neuroinflammation 7 for funding, and Vickie Mewhorter, John Schmeling, and Nancy Peters for antibodies. Neurol Neuroimmunol Neuroinflamm 2015; excellent technical support. 2:e136. doi: 10.1212/NXI.0000000000000136. 9. Gelpi E, Hoftberger R, Graus F, et al. Neuropathological STUDY FUNDING criteria of anti-IgLON5-related tauopathy. Acta Neuropa- Mayo Clinic Center for Individualized Medicine and the Department of thol 2016;132:531–543. Laboratory Medicine and Pathology, Mayo Clinic. 10. Bruggemann N, Wandinger KP, Gaig C, et al. Dystonia, lower limb stiffness, and upward gaze palsy in a patient DISCLOSURE with IgLON5 antibodies. Mov Disord 2016;31:762–764. J.A. Honorat reports no disclosures. L. Komorowski is employed by Euro- 11. Schroder JB, Melzer N, Ruck T, et al. Isolated dysphagia immun AG. K.A. Josephs received research support from NIH/NIA, as initial sign of anti-IgLON5 syndrome. Neurol Neuro- NIH, MINDS, NIH/NIDCD, and The Data Foundation. K. Fechner is immunol Neuroinflamm 2017;4:e302. doi: 10.1212/NXI. employed by Euroimmun AG. E.K. St Louis served on the scientific advisory 0000000000000302. board for Inspire, Inc; is an editorial board member for and guest coeditor of 12. Haitao R, Yingmai Y, Yan H, et al. Chorea and parkinson- Continuum; and received research support from Axovant, Merck, Sunovion ism associated with autoantibodies to IgLON5 and respon- Inc, Mayo Clinic CTSA, and National Center for Advancing Translational sive to immunotherapy. J Neuroimmunol 2016;300:9–10. Science. S.R. Hinson reports no disclosures. S. Lederer is employed by Euro- 13. Gaig C, Graus F, Compta Y, et al. Clinical manifestations of immun Medizinische Labordiagnostika AG. N. Kumar and A. Gadoth – report no disclosures. V.A. Lennon holds a patent for and receives royalties the anti-IgLON5 disease. Neurology 2017;88:1736 1743. from RSR/Kronus for sale of aquaporin-4 autoantibody testing kits and for 14. McKeon A, Pittock SJ, Lennon VA. CSF complements commercial aquaporin-4 autoantibody testing performed outside Mayo serum for evaluating paraneoplastic antibodies and Clinic; received research support from the NIH; has a potential financial NMO-IgG. Neurology 2011;76:1108–1110. interest in the technology “Aquaporin-4 as an aid for cancer diagnosis”; 15. McKeon A, Martinez-Hernandez E, Lancaster E, et al. and receives license fee payments for Non-Mayo sites performing “home Glycine receptor autoimmune spectrum with stiff-man ” brew diagnostic testing for aquaporin-4 autoantibody. S.J. Pittock holds syndrome phenotype. JAMA Neurol 2013;70:44–50. patents for that relate to functional AQP4/NMO-IgG assays and NMO- 16. Dickson DW, Ahmed Z, Algom AA, Tsuboi Y, Josephs IgG as a cancer marker; consulted for Alexion and Medimmune; and KA. Neuropathology of variants of progressive supranu- received research support from Grifols, Medimmune, and Alexion, RO1 clear palsy. Curr Opin Neurol 2010;23:394–400. NS065829-01. A. McKeon has a patent pending for GFAP and MAP1B as markers of neurological autoimmunity and paraneoplastic disorders; con- 17. Crary JF, Trojanowski JQ, Schneider JA, et al. Primary sulted for Grifols, Medimmune, and Euroimmun; and received research age-related tauopathy (PART): a common pathology asso- support from Medimmune and Euroimmun. Go to Neurology.org/nn for ciated with human aging. Acta Neuropathol 2014;128: full disclosure forms. 755–766. 18. Lynch T, Fahn S, Louis ED, Odel JG. Oculofacial-skeletal Received March 7, 2017. Accepted in final form June 8, 2017. myorhythmia in Whipple’s disease. Mov Disord 1997;12: 625–626. REFERENCES 19. Sabater L, Planaguma J, Dalmau J, Graus F. Cellular in- 1. McKeon A, Pittock SJ. Paraneoplastic encephalomyelopa- vestigations with human antibodies associated with the thies: pathology and mechanisms. Acta Neuropathol 2011; anti-IgLON5 syndrome. J Neuroinflammation 2016;13: 122:381–400. 226. 2. Lennon VA, Kryzer TJ, Pittock SJ, Verkman AS, Hinson 20. Pittock SJ, Parisi JE, McKeon A, et al. Paraneoplastic jaw SR. IgG marker of optic-spinal multiple sclerosis binds to dystonia and laryngospasm with antineuronal nuclear the aquaporin-4 water channel. J Exp Med 2005;202: autoantibody type 2 (anti-Ri). Arch Neurol 2010;67: 473–477. 1109–1115. 3. Dalmau J, Tuzun E, Wu HY, et al. Paraneoplastic anti-N- 21. Sanz R, Ferraro GB, Fournier AE. IgLON cell adhesion methyl-D-aspartate receptor encephalitis associated with molecules are shed from the cell surface of cortical neurons ovarian teratoma. Ann Neurol 2007;61:25–36. to promote neuronal growth. J Biol Chem 2015;290: 4. Hutchinson M, Waters P, McHugh J, et al. Progressive 4330–4342. encephalomyelitis, rigidity, and myoclonus: a novel glycine 22. Denton KM, Luff SE, Shweta A, Anderson WP. Differ- receptor antibody. Neurology 2008;71:1291–1292. ential neural control of glomerular ultrafiltration. Clin Exp 5. Tobin WO, Lennon VA, Komorowski L, et al. DPPX Pharmacol Physiol 2004;31:380–386. potassium channel antibody: frequency, clinical accompa- 23. Daniel C, Albrecht H, Ludke A, Hugo C. Nestin expres- niments, and outcomes in 20 patients. Neurology 2014; sion in repopulating mesangial cells promotes their pro- 83:1797–1803. liferation. Lab Invest 2008;88:387–397. 6. Sabater L, Gaig C, Gelpi E, et al. A novel non-rapid-eye 24. Liu Y, Wu J, Wu H, et al. UCH-L1 expression of podo- movement and rapid-eye-movement parasomnia with cytes in diseased glomeruli and in vitro. J Pathol 2009; sleep breathing disorder associated with antibodies to 217:642–653. IgLON5: a case series, characterisation of the antigen, 25. Wang H, Lehtonen S, Chen YC, Heikkila E, Panula P, and post-mortem study. Lancet Neurol 2014;13:575–586. Holthofer H. Neph3 associates with regulation of glomer- 7. Hogl B, Heidbreder A, Santamaria J, Graus F, Poewe W. ular and neural development in zebrafish. Differentiation IgLON5 autoimmunity and abnormal behaviours during 2012;83:38–46. sleep. Lancet 2015;385:1590. 26. Chan KH, Vernino S, Lennon VA. ANNA-3 anti- 8. Simabukuro MM, Sabater L, Adoni T, et al. Sleep disor- neuronal nuclear antibody: marker of lung cancer-related der, chorea, and dementia associated with IgLON5 autoimmunity. Ann Neurol 2001;50:301–311.

8 Neurology: Neuroimmunology & Neuroinflammation Reduced rich-club connectivity is related to disability in primary progressive MS

Jan-Patrick Stellmann, ABSTRACT * MD Objective: To investigate whether the structural connectivity of the brain’s rich-club organization * Sibylle Hodecker, MD is altered in patients with primary progressive MS and whether such changes to this fundamental Bastian Cheng, MD network feature are associated with disability measures. Nadine Wanke, MSc Methods: We recruited 37 patients with primary progressive MS and 21 healthy controls for an Kim Lea Young, MD observational cohort study. Structural connectomes were reconstructed based on diffusion- Claus Hilgetag, PhD weighted imaging data using probabilistic tractography and analyzed with graph theory. Christian Gerloff, MD Christoph Heesen, MD Results: We observed the same topological organization of brain networks in patients and con- Götz Thomalla, MD trols. Consistent with the originally defined rich-club regions, we identified superior frontal, pre- Susanne Siemonsen, MD cuneus, superior parietal, and insular cortex in both hemispheres as rich-club nodes. Connectivity within the rich club was significantly reduced in patients with MS (p 5 0.039). The extent of reduced rich-club connectivity correlated with clinical measurements of mobility (Kendall rank Correspondence to correlation coefficient t 520.20, p 5 0.047), hand function (t 520.26, p 5 0.014), and Dr. Stellmann: information processing speed (t 520.20, p 5 0.049). [email protected] Conclusions: In patients with primary progressive MS, the fundamental organization of the struc- tural connectome in rich-club and peripheral nodes was preserved and did not differ from healthy controls. The proportion of rich-club connections was altered and correlated with disability meas- ures. Thus, the rich-club organization of the brain may be a promising network phenotype for understanding the patterns and mechanisms of neurodegeneration in MS. Neurol Neuroimmunol Neuroinflamm 2017;4:e375; doi: 10.1212/NXI.0000000000000375

GLOSSARY APL 5 average shortest path length; EDSS 5 Expanded Disability Status Scale; FDR 5 false discovery rate; FSL 5 func- tional imaging software library; HC 5 healthy control; MANCOVA 5 multivariate analysis of covariance; MSFC 5 multiple sclerosis functional composite; NHPT 5 Nine-Hole Peg Test; PPMS 5 primary progressive MS; RRMS 5 relapsing-remitting MS; SDMT 5 Symbol Digit Modalities Test; SWI 5 small-world index; T25FW 5 timed 25-foot walk.

MS is the most common autoimmune disease of the CNS, and persistent inflammation as well as chronic progressive neurodegeneration in the brain and spinal cord leads to accumulation of disability.1,2 MRI is currently the best available surrogate marker of MS pathology.3 MRI also permits the use of probabilistic tractography, which allows investigating the integrity of struc- tural connections in the brain. The human brain can be considered as a network and the network’s topology can be studied by graph theory. The network perspective may offer new insights into disease-specific processes such as neurodegeneration and has been applied to several neuropsychiatric diseases.4,5 Recent studies recognized an essential topological feature of the human connectome. The brain network is organized into a so-called rich-club and peripheral nodes.6,7 The rich-club brain regions are

Supplemental data more densely interconnected than expected by chance and form a prominent subnetwork of the at Neurology.org/nn *These authors contributed equally to this work. From the Institut für Neuroimmunologie und Multiple Sklerose (INIMS) (J.-P.S., S.H., N.W., K.L.Y., C.G., C. Heesen, S.S.), Klinik und Poliklinik für Neurologie (J.-P.S., S.H., B.C., N.W., K.L.Y., C. Heesen, G.T.), Institute of Computational Neuroscience (C. Hilgetag), and Department of Diagnostic and Interventional Neuroradiology (S.S.), University Medical Center Hamburg-Eppendorf, Germany. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 brain. They are considered to have a guiding was parcellated into 34 cortical regions per hemisphere and 8 function controlling integration and informa- subcortical regions based on the Destrieux atlas. Subcortical white matter regions corresponding to the 34 cortical regions were used tion flow in the brain network. The rich-club for FSL probabilistic tracking with crossing fibers (probtrackx). architecture assures a highly efficient structural Based on the number of streamlines reaching from one FreeSurfer and functional network organization,8 but region to another, 2 different kinds of networks were defined. might be vulnerable in brain diseases. Altered First, we used the average number of streamlines (forward/ backward) between every 2 regions as edge weight to construct rich-club connectivity has been observed in weighted networks Graw. Second, binary networks (0 5 uncon- schizophrenia,9 migraine,10 or dementia.11 nected and 1 5 connected) were constructed (Gbin) based on MS can also be considered as a network dis- Graw, by discarding connections with a number of streamlines below a given threshold (e.g., 80% of maximum connection order leading to focal and global impairment of strength). This procedure meets the standard to avoid totally 5,12–16 the brain network. However, the rich-club connected networks and reduces the number of edges to the most organization in patients with MS has not been prominent and strongest connections. However, there is no 5,21 analyzed. We aimed to analyze the network consensus, how to define this threshold. We investigated the small-world index (SWI) of individual Gbin at thresholds from topology of structural connectomes of patients 2.5% to 97.5% in steps of 2.5% to determine a suitable common with a predominantly neurodegenerative dis- cutoff for all participants. The best threshold was defined as the ease course. Primary progressive MS (PPMS) highest cutoff with a median SWI above 1 (indicating small- shows less and more diffuse inflammatory dis- world features) but low variance of the SWI (no major bias of this fundamental network feature). For a detailed description, see ease activity than relapsing-remitting MS supplemental material at Neurology.org/nn. (RRMS), and clinical assessment in most cases Graph metrics. Global parameters included strength (Graw), is not blurred by superimposed relapses. There- average shortest path length (APL, Gbin), global efficiency (Gbin), fore, PPMS might be taken as a candidate clustering coefficient (Gbin) and weighted clustering coefficient model for investigating long-term changes of (Graw), and arithmetic mean method. In addition, we computed the node-specific degree (Gbin), betweenness centrality (Gbin), and brain network architecture in MS. We hypoth- strength (Graw). esized that the connectome architecture of pa- tients with PPMS differs from that of healthy Rich club. Rich-club nodes are more densely interconnected than expected by chance (compared with random networks). controls (HCs), in particular with respect to the Thus, the rich club of a given network can be defined by top- rich-club organization, and that these changes ranking nodes based on degree or strength. Eight cortical might be associated with disability measures. regions have been identified as rich-club hubs in healthy in- dividuals, which we used as an a priori–defined rich club: superior frontal, precuneus, superior parietal, and insular cortex in both METHODS Patients and controls. Patients were eligible for hemispheres.6 First, we were interested if node-specific measures this observational cohort study if they were diagnosed with PPMS identify the predefined rich-club nodes in terms of high strength, according to the McDonald criteria 201017 and had an Expanded degree, and betweenness in patients and controls. We then im- Disability Status Scale (EDSS) score of #7.0. Patients (n 5 37) plemented a formal test of the rich-club organization22 and obtained structural MRI and completed a modified multiple scle- investigated whether alternative rich-club definitions (account- rosis functional composite (MSFC) test battery,18 including the ing, e.g., for individual variability) might perform better than the timed 25-foot walk (T25FW, short distance walking speed), the original definition (see supplemental material). We computed the Nine-Hole Peg Test (NHPT, dominant and nondominant fine connectivity within the rich-club, between rich-club and motor hand function), and the Symbol Digit Modalities Test peripheral nodes (so-called feeder connections), and between (SDMT, information processing), which is considered as a sim- peripheral nodes. In addition to the absolute values, rich club, ple-to-administer, less stressful, and valid substitute for the feeder, and peripheral connectivity were each divided by the total MSFC standard Paced Auditory Serial Addition Test.19 We re- connectivity of the individual network. The latter approach was cruited 21 HCs matched by age and sex and applied the same chosen to account for the presumed generalized loss of connec- MRI protocol. tivity in patients with PPMS. The measures were then compared Standard protocol approvals, registrations, and patient by their ability to distinguish between patients with PPMS and consents. Participants were recruited at our MS day hospital HCs, and their association with clinical outcomes and global (2012–2016), provided written informed consent, and the Insti- MRI volumes. tutional Review Board (ethics committee of the Hamburg Cham- Statistics. We performed descriptive statistics according to the ber of physicians, PV3961/PV4405) approved the study. nature of the data as mean with SD or as frequencies and/or per- Image processing and reconstruction of brain networks. centages. Differences between patients and controls were as- Briefly, the MRI protocol included a T1-weighted sequence (0.9 sessed by the Student t test for continuous data and the x2 3 0.9 3 0.9 mm), a T2 sequence (0.5 3 0.5 3 3.0 mm), and test for categorical data. The distribution of connectivity on diffusion tensor imaging (single-shell, 20 directions with rich club, feeder, and peripheral connections was compared noncollinear diffusion gradients [b 5 1,000 s/mm2] and 1 by multivariate analysis of covariance (MANCOVA adjusting nondiffusion-weighted b0 image, 1.9 3 1.9 3 2.0 mm). Images for age, sex, T2 lesion volume, and total connectivity). To were processed with the functional imaging software library (FSL) depict the direction of changes, relative connectivity in the and FreeSurfer software.20 For each participant, the gray matter different compartments was further analyzed by t tests. To

2 Neurology: Neuroimmunology & Neuroinflammation first symptoms was 7.7 years, and a median EDSS of Table 1 Descriptive statistics 3.5 (range 1.5–7) indicated a moderate disability.

PPMS (n 5 39) HCs (n 5 21) p Value Brain volume was lower in patients (p 5 0.003). raw Female/male, n 9/28 9/12 0.242 Stepwise thresholding of G networks increased the SWI of networks only if higher cutoffs were Age, y 52.2 (7.9) 50.4 (6.9) 0.373 applied (figure e-1). Only with a threshold above Disease duration, y 7.6 (5.3) 95%, the median SWI was above borderline values. Brain volume, mm3 1,463,659 (58,775) 1,511,942 (53,783) 0.003 However, we observed in cutoffs above 82.5%, an White matter volume, mm3 764,112 (34,912) 773,561 (34,912) 0.310 increase of the variability and several outliers in pa- Gray matter volume, mm3 699,547 (38,083) 738,380 (35,898) ,0.001a tients and controls. We interpreted this finding as

T1 lesion volume, mm3 4,812 (6,376) artificial noise in the data and, therefore, decided to construct Gbin based on the highest threshold with T2 lesion volume, mm3 6,406 (7,612) acceptable variability, which was 82.5%. EDSS 3.5 (1.5–7)

SDMT 20.7 (1.1) Global graph metrics. Global network metrics are pre- sented in table 2. There was no difference between T25FW, s 6.4 (3.4) patients with PPMS and controls in terms of total con- Abbreviations: PPMS 5 primary progressive MS; HC 5 healthy control (disease duration nectivity (p 5 0.340) or any other graph metric. The since first symptoms, tissue and lesion volumes normalized based on SIENAX results); association between global graph metrics, MRI vol- EDSS 5 Expanded Disability Status Scale; SDMT 5 Symbol Digit Modalities Test (SDs compared with age, sex, and education-matched normalized data); T25FW 5 timed 25-foot umes, and clinical data is summarized in table 3. walk (mean from 2 trials). Although age, T1, and T2 lesion volumes did not cor- Data presented as mean (SD) or median (range). Except from sex (x2 test), group differences relate with graph metrics, strength decreased with longer were compared with the Student t test. disease duration (t 520.30, p 5 0.005), whereas APL account for nonnormal distribution of data, the association increased (t 5 0.23, p 5 0.027) with longer disease between variables was investigated nonparametrically by com- duration. APL correlated inversely with white matter puting 1-sided Kendall correlation coefficient t. Post hoc, we volume (t 520.21, p 5 0.032) and global efficiency compared the predictive value of rich-club connectivity for decreased accordingly (t 5 0.23, p 5 0.025). Walking clinical outcomes with multivariate linear regression. We used t 5 analysis of variance to compare unadjusted models with models speed (T25FW) was associated with strength ( 2 5 5 5 adjusted for brain volume and T2 lesion volume. p Values 0.28, p 0.009) and APL (r 0.28, p 0.011). below 0.05 were considered statistically significant. False dis- The association between strength and disease duration, covery rate (FDR) was used to correct for multiple testing. All respectively, T25FW, and the correlation between APL analyses were performed with Statistics in R 3.2.3, including the and T25FW remained after FDR correction. igraph and tnet packages.23,24 Rich club. A priori–defined rich-club nodes showed RESULTS Cohort and threshold selection. Table 1 the highest betweenness of all nodes (figure 1), which summarizes descriptive statistics. The mean age of indicates that their role as an important junction patients with PPMS was 52.2 years and did not differ within the networks was preserved. These nodes also from HCs (50.4, p 5 0.282). Disease duration since ranked within the top 9 nodes based on strength and

Table 2 Graph metrics: patients and controls

PPMS vs PPMS (n 5 39) HCs (n 5 21) HCs

Graw Gbin Graw Gbin Graw Gbin

Graph strength 2,625,250,401 2,463,142,190 2,660,851,915 2,494,621,338 0.764 0.788 (506,437,884) (499,636,345) (383,876,086) (378,863,703)

Average shortest 1.03 (0.02) 2.17 (0.1) 1.03 (0.02) 2.16 (0.07) 0.876 0.438 path length

Global efficiency 2.23 (0.02) 1.1 (0.07) 2.23 (0.02) 1.12 (0.07) 0.876 0.130

Clustering 0.97 (0.01) 0.51 (0.02) 0.97 (0.01) 0.51 (0.01) 0.826 0.058 coefficient

Clustering 0.99 (0.01) 0.61 (0.02) 0.99 (0) 0.60 (0.02) 0.838 0.051 coefficient weighted

Small-world 2.12 (0.03) 2.04 (0.14) 2.12 (0.03) 2.00 (0.09) 0.800 0.210 index

Abbreviations: Gbin 5 binary networks; Graw 5 weighted networks; HC 5 healthy control; PPMS 5 primary progressive MS. Data presented as mean (SD). Comparison of cohorts by the Student t test.

Neurology: Neuroimmunology & Neuroinflammation 3 Table 3 Association of global graph metrics, MRI volumes, and clinical data in PPMS

Average shortest Weighted Strength path length Efficiency Clustering clustering

Age 20.13 (0.140) 0.01 (0.484) 20.08 (0.261) 0.03 (0.602) 20.03 (0.408)

Disease duration since 20.30 (0.005)a,b 0.23 (0.027)a 20.17 (0.077) 20.12 (0.158) 20.11 (0.178) diagnosis

Brain volume 0.06 (0.297) 20.14 (0.118) 0.08 (0.254) 20.12 (0.860) 20.12 (0.848)

White matter volume 0.15 (0.098) 20.21 (0.032)a 0.23 (0.025)a 20.1 (0.815) 20.11 (0.829)

Gray matter volume 0.01 (0.495) 20.05 (0.344) 20.05 (0.666) 20.11 (0.835) 20.12 (0.860)

T1 hypointense lesion 0.02 (0.567) 0.14 (0.123) 20.06 (0.297) 0.13 (0.877) 0.18 (0.943) volume

T2 hyperintense lesion 20.05 (0.344) 0.19 (0.051) 20.08 (0.262) 0.02 (0.567) 0.05 (0.666) volume

EDSS 20.05 (0.349) 0.09 (0.231) 20.04 (0.359) 0.08 (0.744) 0.06 (0.680)

T25FW 20.28 (0.009)a,b 0.28 (0.011)a,b 20.13 (0.146) 20.01 (0.482) 20.13 (0.140)

SDMT 20.08 (0.738) 20.16 (0.088) 20.04 (0.635) 20.15 (0.884) 20.15 (0.889)

NHPT dominant hand 0.03 (0.593) 0.13 (0.136) 20.09 (0.216) 0.04 (0.633) 0.08 (0.743)

NHPT nondominant hand 0.19 (0.948) 0.02 (0.43) 0.14 (0.887) 0.19 (0.945) 0.26 (0.987)

Abbreviations: EDSS 5 Expanded Disability Status Scale; NHPT 5 Nine-Hole Peg Test; PPMS 5 primary progressive MS; SDMT 5 Symbol Digit Modalities Test; T25FW 5 timed 25-foot walk. Associations measured with Kendall t, p values in brackets. a p Values below 0.05. b False discovery rate–corrected p values below 0.05.

degree (figures e-2 and e-3). Overall, the large-scale peripheral connectivity were associated with increas- organization of the connectomes did not differ ing T2 lesion load (t 520.21, p 5 0.034) and lower between patients with PPMS and HCs. The order of gray matter volumes (t 5 0.21, p 5 0.037). Rela- nodes was highly correlated between patients and tively lower rich-club connectivity was associated with controls, based on strength (r 5 0.99, p 5 0.001), NHPT (nondominant: t 520.26, p 5 0.014) and degree (r 5 0.99, p , 0.001), and betweenness (r 5 T25FW (t 520.20, p 5 0.047). An increased rel- 0.92, p , 0.001). Formal testing confirmed the rich- ative peripheral connectivity was instead linked to club organization in average connectomes of patients lower cognitive performance on the SDMT (t 5 and controls (figure e-4). 20.20, p 5 0.049). FDR-corrected p values re- For the a priori–defined rich club, absolute con- mained significant for NHPT and relative rich-club nectivity strength was lower in patients with PPMS connectivity. Post hoc, an adjustment for brain vol- than that in controls (p 5 0.038, FDR-corrected p 5 ume and T2 lesions did not improve the predictive 0.114). The absolute strength of feeder and periph- value of rich-club connectivity for T25FW perfor- eral connectivity did not differ (p 5 0.317 and p 5 mance (p 5 0.283). Concerning NHPT and SDMT, 0.474, FDR-corrected both p 5 0.474). Corrected the adjustment for brain volume and T2 lesions per- for total connectivity, T2 lesion volume, age, and sex, formed better than the simple models (both p , we observed an altered distribution of connectivity 0.001). Alternative rich-club definitions did not show between rich club, feeder, and periphery in patients a better ability to discriminate between patients with with PPMS (MANCOVA p 5 0.011): pairwise T PPMS and HCs, nor were they more closely associ- tests revealed that the percentage of connections ated with disability (table e-1). within the rich-club, that is, the relative rich-club con- nectivity, was reduced in patients (p 5 0.013, FDR- DISCUSSION The rich-club organization of the corrected p 5 0.039), whereas peripheral connectivity human connectome has been identified as a funda- was relatively increased in patients than that in con- mental feature of brain networks.6 Here, we investi- trols (p 5 0.040, FDR-corrected p 5 0.060, figure 2, gated for the first time how the rich-club organization A–C). of the human structural connectome is affected in pa- Absolute rich-club connectivity as well as propor- tients with PPMS. Generally, we observed a preserved tional distribution of connections between the rich rich-club organization in patients with PPMS. How- club and periphery was not associated with sex, age, ever, compared with HCs, the connectivity within the disease duration, or T1 lesion volume. By contrast, rich club was reduced. In addition, lower rich-club relatively lower rich-club connectivity and higher connectivity was associated with higher disability.

4 Neurology: Neuroimmunology & Neuroinflammation Figure 1 Node-specific graph metrics: Betweenness

Boxplots ordered by median values of nodes of Gbin. (A) Healthy controls and (B) primary progressive MS.

Based on the weighted rich-club effect, we con- did not differ from those in moderately disabled pa- firmed the rich-club organization in controls and pa- tients with PPMS. Although the global connectivity tients. The nodes forming the rich club in controls was lower in patients, the fundamental organizational

Neurology: Neuroimmunology & Neuroinflammation 5 Figure 2 Rich-club connectivity

Schematic representation of the rich-club organization in average primary progressive MS (PPMS) connectomes (top right). Boxplots show differences between PPMS (blue) and healthy controls (HCs) (orange) in relative connectivity within the rich club (A), for feeders (B), and within the periphery (C). Dotplots show association between relative rich-club connectivity and (D) T25FW 5 timed 25-foot walk, (E) SDMT 5 Symbol Digit Modalities Test, (F) NHPT 5 Nine-Hole Peg Test dominant hand, and (G) NHPT nondominant hand. Lines and colored areas represent regression estimates and their confident intervals. For details, see Methods and Results sections.

principles did not seem to be affected. Superior fron- density of rich-club connections compared with con- tal, precuneus, superior parietal, and insular cortex in trols, whereas peripheral connectivity was relatively both hemispheres formed the rich club in patients increased, which might be due to a lower extent of and controls in line with the originally defined rich- connectivity loss in the periphery compared with club nodes.6 The dominance of these nodes was con- the rich-club or might result from compensatory sistent across different nodal graph metrics. peripheral rewiring. Our adjustment strategy demon- The absolute number of rich-club connections strates that the observed loss of rich-club connections was lower in patients than that in controls, whereas cannot be explained solely by the global loss of con- peripheral and feeder connections did not differ. nectivity but indicates a disease-specific pattern. Adjusted for the total connectivity of individual Here, reduced rich-club connectivity was associated brains, we observed an altered distribution of connec- with T2 lesion load and gray matter volume, connect- tions between the rich club and periphery. Specifi- ing our findings to previous observations.15 More- cally, in patients with PPMS, we observed a lower over, we observed a correlation of the change in

6 Neurology: Neuroimmunology & Neuroinflammation rich-club connectivity with clinical measures of large-scale functional networks.31 The default mode mobility, hand function, and cognition in our network is considered to represent a backbone of patients. structural-functional organization and shows a hetero- Alterations to the distribution of connections geneous pattern of activation and deactivation in between rich-club and peripheral nodes have also MS.32 Altered default mode connectivity contrasts been observed in other neuropsychiatric diseases the connectivity in more peripheral networks, adding and support a disease specificity of distinctive pat- further evidence for inverse effects in hubs and the terns. In migraine patients, an increased number of periphery.31 However, it remains an unresolved issue feeder connections is associated with a higher network how reorganization of functional connectivity needs efficiency and suspected to cause a higher integration to be interpreted and what the underlying pathologic of subnetworks involved in pain processing.10 How- or repair mechanisms are. Moreover, it is unknown ever, patients with schizophrenia show reduced rich- how these functional changes translate into altered club connectivity,25 and cognitive decline and structural connectivity. The cross-sectional nature of reduced general function are associated with a pro- our study does not allow determining when the shift nounced loss of rich-club connections over 3 years.9 from rich-club to peripheral connectivity occurs. Predominantly neurodegenerative diseases such as However, preliminary data in RRMS indicate that Parkinson disease seem to be associated with a pro- the shift toward more peripheral connectivity is nounced loss of peripheral and feeder connec- detectable within a time frame of 6 months and might tions.11,26 Within this context, our findings indicate be reduced by an exercise intervention.33 A cross- that diffuse perturbations of brain connectivity in sectional study in RRMS found also an increased PPMS result in a pronounced loss of connectivity structural connectivity on a local level.34 Taking the within the rich club, and that the overall pattern of functional data into account, one might hypothesize connectivity loss differs from primarily neurodegen- that a continuous loss of rich-club connectivity can erative diseases. However, this cross-sectional study only be compensated by peripheral connections in the does not allow to distinguish between the impact of early phase. inflammatory and neurodegenerative mechanisms on In contrast to our rich-club analyses, conventional the assessed connectivity loss. global graph metrics did not differ between patients This interpretation is supported by data indicating and controls. The insensitivity of global graph metrics that atrophy appears to follow nonrandom patterns in has been described before. For example, they were not MS correlated with cognitive impairment and disabil- able to separate comatose patients from HCs in an ity.12 Most of the patterns include the insula, which fMRI study, while hub regions showed a clear loss belongs to the rich-club regions. Moreover, a close in connectivity.35 These findings support our inter- correlation between T2 lesions and reduced local effi- pretation that PPMS does not lead to major topolog- ciency of the insula has been described before,15 and ical changes but rather to subtle alterations of the the affection of hub regions in early RRMS has brain network. recently been confirmed in structural and functional Our findings are of explorative nature and need connectomes.27 These findings support our observa- further validation. The sample size is relatively small, tion that PPMS affects particularly the regions of the but in line with similar studies investigating rich-club brain with a prominent and integrative role in the connectivity.10,11,26 The study was sufficiently pow- structural connectome, whereas peripheral connec- ered to detect alterations in the rich-club connectiv- tions are less compromised, preserved, or even re- ity, but was too small to investigate the association wired due to neurorepair. A longitudinal study of with other MRI metrics in depth. For example, we fMRI connectivity found a compensatory upregula- did not detect a difference in white matter volume tion in early disease stages without relevant disabil- between patients and controls. Moreover, we could ity.28 A global loss of connectivity was observed in not apply model selection in multivariate statistics to more disabled patients and correlated with accumu- determine the specificity of rich-club alterations for lation of disability. Cross-sectional studies support disability in comparison with other MRI measure- the observation that functional reorganization in ments. Concerning the mechanisms behind our ob- MS is complex and leads to different patterns of acti- servations, our interpretations remain hypothetical, as vation and deactivation related to disability.29 How- we did not investigate longitudinal changes. Not ever, the benefit of this reorganization is including relapsing-remitting patients limits the gen- questionable.30 Recruitment of additional neural eralization of our findings, but we were primarily resource might be inefficient and rather maladaptive interested in neurodegenerative aspects of MS and than adaptive.30 thus aimed to restrict the influence of acute inflam- The topology of network reorganization can as matory disease activity as much as possible by study- well be deduced by contrasting alterations in the ing a PPMS cohort. However, especially spinal cord

Neurology: Neuroimmunology & Neuroinflammation 7 pathology in this patient group might influence dis- Werner-Otto-Foundation. C. Heesen received speaker honoraria from ability by means of the EDSS and mobility assess- Biogen, Merck, Genzyme, and Novartis; is on the editorial board for International Journal of MS Care; and received research support from ment and make the interpretation of our association Genzyme, Sanofi Aventis, Biogen, Novartis, Roche, Merck, and German results more difficult. Furthermore, the applied Ministry of Research. G. Thomalla served on the scientific advisory board method to reconstruct connectomes shares some for TEA Stroke Trial; received travel funding and/or speaker honoraria from Bayer, Boehringer Ingelheim, Daiichi Sankyo, and Bristol-Myers inherent limitations. Although the method is well Squibb/Pfizer; consulted for Acandis and GlaxoSmithKline; and received accepted, it may be affected by methodological and research support from the European Union, Deutsche Forschungsgesell- disease-specific aspects, such as the undistinguishable, schaft, and Corona Foundation. S. Siemonsen reports no disclosures. Go but heterogenic histopathology of MS lesions.36,37 MS to Neurology.org/nn for full disclosure forms. lesions affect the tractography results such as they Received December 12, 2016. Accepted in final form May 17, 2017. may underestimate true structural connectivity.38 However, false-negative connections are known to REFERENCES have a lower impact on the network topology than 1. Dendrou CA, Fugger L, Friese MA. Immunopathology of false-positive connections,36 and in our data set, the multiple sclerosis. Nat Rev Immunol 2015;15:545–558. global connectivity did not differ between patients 2. Hauser SL, Chan JR, Oksenberg JR. Multiple sclerosis: and controls. Our analysis followed the original anal- prospects and promise. Ann Neurol 2013;74:317–327. ysis of the rich club that restricted the topological 3. Rovira À, Wattjes MP, Tintoré M, et al. Evidence-based 6,25 guidelines: MAGNIMS consensus guidelines on the use of analyses to cortical regions. Thus, we cannot con- MRI in multiple sclerosis—clinical implementation in the clude about the role of subcortical gray matter or the diagnostic process. Nat Rev Neurol 2015;11:471–482. cerebellum in the rich-club topology. 4. Kaiser M. The potential of the human connectome as In conclusion, PPMS seems to induce a loss of a biomarker of brain disease. Front Hum Neurosci structural connections between important brain re- 2013;7:484. gions. Our findings emphasize to interpret the 5. Griffa A, Baumann PS, Thiran JP, Hagmann P. Structural connectomics in brain diseases. Neuroimage 2013;80: impairment of rich-club connections as disease spe- 515–526. cific and disability related. Thus, rich-club connectiv- 6. van den Heuvel MP, Sporns O. Rich-club organization of the ity is a promising indicator for understanding and human connectome. J Neurosci 2011;31:15775–15786. monitoring the patterns and mechanisms of neurode- 7. Cao M, He Y, Dai Z, et al. Early development of func- generation in PPMS. tional network segregation revealed by connectomic anal- ysis of the preterm human brain. Cereb Cortex 2016;27: AUTHOR CONTRIBUTIONS 1949–1963. Study concept and design: J.-P.S., S.H., B.C., C. Heesen, G.T., and S.S. 8. de Reus MA, van den Heuvel MP. The parcellation-based Acquisition, analysis, or interpretation of data: J.-P.S., S.H., B.C., N.W., connectome: limitations and extensions. Neuroimage K.L.Y., C. Hilgetag, C.G., C. Heesen, G.T., and S.S. Drafting of the 2013;80:397–404. manuscript: J.-P.S. and S.H. Critical revision of the manuscript 9. Collin G, de Nijs J, Hulshoff Pol HE, Cahn W, van den for important intellectual content: J.-P.S., S.H., B.C., N.W., K.L.Y., Heuvel MP. Connectome organization is related to longitu- C. Hilgetag, C.G., C. Heesen, G.T., and S.S. Statistical analysis: J.-P.S., dinal changes in general functioning, symptoms and IQ in S.H., B.H., G.T., and S.S. Study supervision: J.-P.S., K.L.Y., S.H., chronic schizophrenia. Schizophr Res 2016;173:166–173. and C. Heesen. 10. Li K, Liu L, Yin Q, et al. Abnormal rich club organization and impaired correlation between structural and functional STUDY FUNDING connectivity in migraine sufferers. Brain Imaging Behav Data collection was partially supported by grants from Merck Serono and 2017;11:526–540. Novartis. The German Ministry for Education and Research supported 11. Daianu M, Mezher A, Mendez MF, Jahanshad N, Jimenez the work (BMBF, SFB 936/A1, C1, C2, Z3, and TRR 169/A2). EE, Thompson PM. Disrupted rich club network in behav- ioral variant frontotemporal dementia and early-onset Alz- DISCLOSURE heimer’s disease. Hum Brain Mapp 2016;37:868–883. J.-P. Stellmann served on the scientific advisory board for Genzyme; 12. Steenwijk MD, Geurts JJG, Daams M, et al. Cortical received travel funding and/or speaker honoraria from Genzyme, Biogen, and Novartis; and received research support from Merck Serono and atrophy patterns in multiple sclerosis are non-random – Biogen. S. Hodecker received travel funding from Genzyme and Roche. and clinically relevant. Brain 2016;139:115 126. B. Cheng, N. Wanke, K.L. Young, and C. Hilgetag report no disclosures 13. Tewarie P, Schoonheim MM, Schouten DI, et al. Func- and received research support from DFG SFB and DFG TRR. C. Gerloff tional brain networks: linking thalamic atrophy to clinical served on the scientific advisory board for Bayer Vita, Boehringer Ingel- disability in multiple sclerosis, a multimodal fMRI and heim, EBS Technologies, Silk Road Medical, Acticor Biotech, Amgen, MEG Study. Hum Brain Mapp 2015;36:603–618. and Prediction Bioscience; received travel funding and/or speaker hono- 14. Li Y, Jewells V, Kim M, et al. Diffusion tensor imaging raria from Bayer Vital, Boehringer Ingelheim, Biogen, ev3/Covidien, based network analysis detects alterations of neuroconnec- GlaxoSmithKline, Grifols, Inomed, Lundbeck, Nexstim, Pfizer, Sanofi tivity in patients with clinically early relapsing-remitting Aventis, UCB, and Merck Serono; served on the editorial board for multiple sclerosis. Hum Brain Mapp 2013;34:3376–3391. INFO Neurologie Psychiatrie and Aktuelle Neurologie; was editor of the textbook “Therapie und Verlauf Neurologischer Erkrankungen”; con- 15. He Y, Dagher A, Chen Z, et al. Impaired small-world sulted for EBS Technologies; and received research support from Merz, efficiency in structural cortical networks in multiple scle- Novartis, NeuroConn, DFG, BMBF/DFG, EU, DFG: TRR, Wegener rosis associated with white matter lesion load. Brain 2009; Foundation, Movement Disorders Science, Schilling Foundation, and 132:3366–3379.

8 Neurology: Neuroimmunology & Neuroinflammation 16. Shu N, Liu Y, Li K, et al. Diffusion tensor tractography 28. Faivre A, Robinet E, Guye M, et al. Depletion of brain reveals disrupted topological efficiency in white matter functional connectivity enhancement leads to disability structural networks in multiple sclerosis. Cereb Cortex progression in multiple sclerosis: a longitudinal resting- 2011;21:2565–2577. state fMRI study. Mult Scler J 2016;22:1695–1708. 17. Polman CH, Reingold SC, Banwell B, et al. Diagnostic 29. Rocca MA, Valsasina P, Meani A, Falini A, Comi G, criteria for multiple sclerosis: 2010 revisions to the Filippi M. Impaired functional integration in multiple “McDonald criteria.” Ann Neurol 2011;69:292–302. sclerosis: a graph theory study. Brain Struct Funct 2016; 18. Fischer JS, Rudick RA, Cutter GR, Reingold SC. The 221:115–131. multiple sclerosis functional composite measure (MSFC): 30. Enzinger C, Pinter D, Rocca MA, et al. Longitudinal an integrated approach to MS clinical outcome assessment. fMRI studies: exploring brain plasticity and repair in Mult Scler 1999;5:244–250. MS. Mult Scler J 2016;22:269–278. 19. Drake A, Weinstock-Guttman B, Morrow S, Hojnacki D, 31. Rocca MA, Valsasina P, Martinelli V, et al. Large-scale Munschauer F, Benedict R. Psychometrics and normative neuronal network dysfunction in relapsing-remitting mul- data for the multiple sclerosis functional composite: replac- tiple sclerosis. Neurology 2012;79:1449–1458. ing the PASAT with the symbol digit modalities test. Mult 32. Zhou F, Zhuang Y, Gong H, et al. Altered inter-subregion Scler 2010;16:228–237. connectivity of the default mode network in relapsing 20. Fischl B. FreeSurfer. Neuroimage 2012;62:774–781. remitting multiple sclerosis: a functional and structural 21. Welton T, Kent DA, Auer DP, Dineen RA. Reproducibil- connectivity study. PLoS One 2014;9:e101198. ity of graph-theoretic brain network metrics: a systematic 33. Hodecker SC, Siemonsen S, Kjølhede T, et al. Physical review. Brain Connect 2015;5:193–202. exercise-related changes in structural connectome architec- 22. Opsahl T, Colizza V, Panzarasa P, Ramasco JJ. Promi- ture in patients with relapsing-remitting multiple sclerosis. nence and control: the weighted rich-club effect. Phys Mult Scler J 2016;22:35. Rev Lett 2008;101:1–4. 34. Fleischer V, Gröger A, Koirala N, et al. Increased struc- 23. Opsahl T. Structure and Evolution of Weighted Networks. tural white and grey matter network connectivity compen- London: University of London (Queen Mary College), sates for functional decline in early multiple sclerosis. Mult London, UK; 2009. Scler 2017;23:432–441. 24. Csardi G, Nepusz T. The igraph software package for 35. Achard S, Delon-Martin C, Vértes PE, et al. Hubs of complex network research. Inter J Complex Syst 2006; brain functional networks are radically reorganized in 1695:1695. comatose patients. Proc Natl Acad Sci USA 2012;109: 25. Collin G, Kahn RS, De Reus MA, Cahn W, Van Den 20608–20613. Heuvel MP. Impaired rich club connectivity in unaffected 36. Zalesky A, Fornito A, Cocchi L, Gollo LL, van den siblings of schizophrenia patients. Schizophr Bull 2014;40: Heuvel MP, Breakspear M. Connectome sensitivity or 438–448. specificity: which is more important? Neuroimage 26. Li C, Huang B, Zhang R, et al. Impaired topological 2016;142:407–420. architecture of brain structural networks in idiopathic Par- 37. Bastiani M, Shah NJ, Goebel R, Roebroeck A. Human kinson’s disease: a DTI study. Brain Imaging Behav 2016; cortical connectome reconstruction from diffusion 22:1–16. weighted MRI: the effect of tractography algorithm. 27. Shu N, Duan Y, Xia M, et al. Disrupted topological orga- Neuroimage 2012;62:1732–1749. nization of structural and functional brain connectomes in 38. Droby A, Fleischer V, Carnini M, et al. The impact of clinically isolated syndrome and multiple sclerosis. Sci Rep isolated lesions on white-matter fiber tracts in multiple 2016;6:29383. sclerosis patients. Neuroimage Clin 2015;8:110–116.

Neurology: Neuroimmunology & Neuroinflammation 9 Lipoic acid in secondary progressive MS A randomized controlled pilot trial

Rebecca Spain, MD, ABSTRACT MSPH Objective: To determine whether lipoic acid (LA), an endogenously produced antioxidant, slowed Katherine Powers, BA, the whole-brain atrophy rate and was safe in secondary progressive MS (SPMS). CCRP Methods: Patients with SPMS aged 40–70 years enrolled in a single center, 2-year, double-blind, Charles Murchison, MS randomized trial of daily oral 1,200 mg LA vs placebo. Primary outcome was change in annualized Elizabeth Heriza, MD percent change brain volume (PCBV). Secondary outcomes were changes in rates of atrophy of Kimberly Winges, MD segmented brain, spinal cord, and retinal substructures, disability, quality of life, and safety. Vijayshree Yadav, MD, Intention-to-treat analysis used linear mixed models. MSCR Michelle Cameron, MD, Results: Participation occurred between May 2, 2011, and August 14, 2015. Study arms of LA PT (n 5 27) and placebo (n 5 24) were matched with mean age of 58.5 (SD 5.9) years, 61% women, Ed Kim, MD mean disease duration of 29.6 (SD 9.5) years, and median Expanded Disability Status Score of Fay Horak, PhD, PT 6.0 (interquartile range 1.75). After 2 years, the annualized PCBV was significantly less in the LA Jack Simon, MD, PhD arm compared with placebo (20.21 [standard error of the coefficient estimate (SEE) 0.14] vs Dennis Bourdette, MD 20.65 [SEE 0.10], 95% confidence interval [CI] 0.157–0.727, p 5 0.002). Improved Timed 25- Foot Walk was almost but not significantly better in the LA than in the control group (20.535 [SEE 0.358] vs 0.137 [SEE 0.247], 95% CI 21.37 to 0.03, p 5 0.06). Significantly more Correspondence to gastrointestinal upset and fewer falls occurred in LA patients. Unexpected renal failure (n 5 1) Dr. Spain: and glomerulonephritis (n 5 1) occurred in the LA cohort. Compliance, measured by pill counts, [email protected] was 87%. Conclusions: LA demonstrated a 68% reduction in annualized PCBV and suggested a clinical benefit in SPMS while maintaining favorable safety, tolerability, and compliance over 2 years. ClinicalTrials.gov identifier: NCT01188811. Classification of evidence: This study provides Class I evidence that for patients with SPMS, LA reduces the rate of brain atrophy. Neurol Neuroimmunol Neuroinflamm 2017;4:e374; doi: 10.1212/ NXI.0000000000000374

GLOSSARY AE 5 adverse event; CI 5 confidence interval; EAE 5 experimental autoimmune encephalomyelitis; EDSS 5 Expanded Disability Status Scale; GI 5 gastrointestinal; ITT 5 intention to treat; LA 5 lipoic acid; MP-RAGE 5 magnetization-prepared rapid acquisition gradient echo; OCT 5 optical coherence tomography; PCBV 5 percent change brain volume; PI 5 principal investigator; RRMS 5 relapsing-remitting MS; SAE 5 serious AE; SEE 5 standard errors of the coefficient estimate; SPMS 5 secondary progressive MS.

By 2 decades, the majority with relapsing-remitting MS (RRMS) have secondary progressive MS (SPMS). SPMS pathophysiology likely involves mitochondrial dysfunction, microglial activa- tion, vascular endothelial disruption, and effects of meningeal lymphoid-like tissues.1 The resulting neurodegeneration and accelerated brain atrophy correlate with functional disability; thus whole-brain atrophy is the current gold-standard MRI surrogate outcome measure for

Supplemental data at Neurology.org/nn From the Neurology Division (R.S., V.Y., M.C., E.K., D.B.), Research Service (K.P., E.H.), and Department of Ophthalmology (K.W.), Veterans Affairs Portland Health Care System, OR; and Department of Neurology (R.S., C.M., K.W., V.Y., E.K., F.H., J.S., D.B.), Advanced Imaging Research Center (K.P.), and Casey Eye Institute (K.W.), Oregon Health & Science University, Portland. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 SPMS trials.2 Targeting specific pathophysio- glatiramer acetate or b-interferon during the study. Exclusion logic processes is a rational strategy for treating criteria were use of natalizumab, immunosuppressants, chemo- therapies, or scheduled IV corticosteroid treatments within 1 year SPMS. of enrollment, corticosteroid treatment for relapse within 60 days Lipoic acid (LA) is an endogenously pro- of enrollment, LA within 30 days of enrollment, MRI constraints, duced antioxidant with multiple biological self-reported ocular disease that could confound optical coherence functions including free-radical scavenging, tomography (OCT) interpretation, pregnant or breastfeeding, significant active concurrent illness, uncontrolled or insulin- metallic ion chelation, regeneration of intra- dependent diabetes, and lack of English fluency. Because of cellular glutathione, and oxidative damage slow recruitment during the first 8 months, the EDSS limit of 6.0 repair of macromolecules.3 In mitochondria, was eliminated. the LA/dihydrolipoic acid redox couple is Standard protocol approvals, registrations, and patient a key cofactor for the pyruvate dehydrogenase consents. The study was approved by VAPORHCS and OHSU Institutional Review Boards. Written consent was obtained from complex of oxidative respiration and aids nu- all the participants. The study was registered on ClinicalTrials.gov 4 cleic acid synthesis. LA modulates the PKB/ (NCT01188811) and conducted following 2010 CONSORT Akt signaling pathway important for vascular guidelines.14 endothelial integrity, affects transcription fac- Study personnel roles. The PI conducted screening visits and tor Nrf2, and acts as an insulin mimetic.5,6 initial EDSS examinations for randomization, evaluated adverse Our center and others have shown that LA events (AEs), and served as the study monitor. Blinded neurolo- gists and a neurology advanced practitioner served as EDSS exam- reduced disability in experimental autoim- iners. Blinded study coordinators collected mobility data, mune encephalomyelitis (EAE), diminished questionnaires, and maintained databases. inflammatory cell migration into spinal cords Investigational drug. An Investigational New Drug indication and optic nerves, and inhibited macrophage/ was obtained by the PI (no. 110132). Pure Encapsulations microglial activation.7,8 Oral ingestion of LA (Sudbury, MA) provided gelatin capsules containing 600 mg race- mic LA or placebo. Placebo capsules contained Avicel (microcellu- by patients with MS produced blood levels lose crystal) and 4.3 mg of quercetin (a bioflavonoid) that rendered 9 comparable with those in EAE. In clinical the placebo a yellow color, similar to LA. Expiration of the study trials, LA is well tolerated; common adverse drug was extended once during the study by Pure Encapsulations reactions were gastrointestinal (GI) intoler- after retesting of sample capsules determined continued stability. ance, headache, malodorous urine, and Study timeline. Screen and baseline visits were #30 days apart. rash.10,11 Herein, we report results of a 2- Subsequent visits at months 3, 6, 12, 18, and 24 occurred 62 weeks. A month was defined as 4 weeks for scheduling purposes. year, randomized controlled trial to determine MRI and OCT occurred at baseline, months 12 and 24. Clinical whether LA reduced rates of whole-brain atro- outcome measures were collected at baseline and every 6 months. phy, slowed clinical deterioration, and was safe Safety laboratory measures were performed at each visit. Tele- in SPMS. phone calls occurred between visits and after study completion to capture AEs.

Outcomes. Primary outcome was difference in annual percent METHODS Study design. This was a prospective, single-site, change brain volume (PCBV) on MRI by Structural Image Eval- 2-year, phase II, double-blind, randomized, placebo-controlled uation, using Normalisation, of Atrophy (SIENA). Secondary trial of 1,200 mg daily oral racemic LA to answer the following outcomes included atrophy rates of segmented brain, spinal cord, primary research question with Class I level of evidence: would and retinal substructures, changes in disability, quality of life, and LA reduce the rate of whole-brain atrophy in SPMS? Secondary safety. research questions were to determine whether LA would reduce rates of atrophy of segmented brain, spinal cord, and retinal MRI acquisition protocol. MRI acquisition and analyses substructures, reduce deterioration of disability and quality of life, details are shown in appendix e-1 at Neurology.org/nn. The and be safe in SPMS. Recruitment occurred between May 2011 following sequences were acquired using a Philips Achieva 3.0T and October 2013, with last visit being in August 2015. The X-series with Quasar Dual gradient systems: (1) 3D high- study was conducted at the Veterans Affairs Portland Health Care resolution magnetization-prepared rapid acquisition gradient System (VAPORHCS), Portland, OR, with some procedures at echo (3D MP-RAGE) with 1 mm3 voxels for high-resolution Oregon Health & Science University (OHSU), Portland, OR. structural (T1-weighted) information. The upper cervical spinal cord was intentionally included in the series through positioning; Participants. Inclusion criteria were ages 40–70 years, prior (2) 3D fluid-attenuated inversion recovery (3D FLAIR) series RRMS (2005 McDonald criteria), and current SPMS defined with 1 mm3 voxels; (3) conventional 3 mm (0.3 gap) axial 2D by MS disability progression in the absence of clinical relapse proton density/T2-weighted sequences with in-plane resolution during the prior 5 years as determined by the principal investi- 1mm2; and (4) 3-mm sagittal 2D proton density/T2-weighted gator (PI) based on history and chart review.12 Progression was spinal cord sequences. No intravascular contrast was used. defined as sufficient to change a Functional System on the Expanded Disability Status Scale (EDSS) or effect a meaningful MRI analyses. MRIs were reviewed by a neuroradiologist for functional change (e.g., stopped working due to cognitive unexpected findings. A single trained MRI analyst performed decline).13 Participants were permitted to start, stop, or continue lesion counts, volumetrics, and cortical thickness analyses. The

2 Neurology: Neuroimmunology & Neuroinflammation Figure 1 2010 CONSORT flow diagram

PI conducted spinal cord cross-sectional thickness and lesion protocols, respectively. OCTs were reviewed by an experienced occupancy analyses. Both were directed by the study neuroradi- neuro-ophthalmologist. Excluded scans had confounding findings, ologist. Cerebral T2-hyperintense lesion volumes and maps were artifact, misalignment, or signal strength less than 7. obtained using Lesion TOpology-preserving Anatomical Seg- Clinical measures. Disability was captured by the EDSS. The mentation (Lesion-TOADS).15 FSL tools were used to create same EDSS examiner was used for a given participant throughout lesion-filled MP-RAGE images. SIENAX was used to determine the study to the extent possible. Mobility measures for ambula- cross-sectional whole-brain, white, and grey matter volumes.16 tory participants were Timed 25-Foot Walk (T25FW), Multi- Whole-brain atrophy was determined using SIENA from the FSL ple Sclerosis Walking Scale (MSWS-12) Questionnaire, and package.16 Subcortical deep grey matter volumes were measured Activities-specific Balance Confidence (ABC) Questionnaire.18–20 using FIRST.16 Cortical thickness, cortical volumetric segmen- Symbol Digit Modalities Test (SDMT) tested cognition.21 tation, and analyses of these were performed with FreeSurfer RAND 36-Item Short Form Health Survey assessed quality of (surfer.nmr.mgh.harvard.edu/) and its longitudinal processing life.22 stream.17 Spinal cord cross-sectional area was recorded at C1. An in-house estimation of the relative percent of the spinal cord from Safety monitoring. AEs were categorized using the Common the foramen magnum to the lower edge of C7 occupied by MS Terminology Criteria for Adverse Events version 4.0 (CTCAE lesions was devised for descriptive purposes. MRIs were graded v4.0). Unscheduled visits occurred for relapses, AEs, and early for quality (good, fair, poor, and unusable). Poor and unusable study termination visits. Safety monitoring laboratory tests (com- scans were excluded from the analyses. plete blood count and kidney and liver panels) were checked at each visit. The Columbia-Suicide Severity Rating Scale OCT. Participants underwent spectral domain OCT (Cirrus HD- (C-SSRS) was administered at every visit. A 3-member data safety OCT; Carl Zeiss Meditec, Inc., Dublin, CA) in each eye after and monitoring board met every year. pharmacologic dilation (1% tropicamide and 0.5% proparacaine hydrochloride). Peripapillary and macular scans were obtained with Sample size, randomization, and blinding. The study was Optic Disc Cube 200 3 200 and Macular Cube 512 3 128 powered for comparison of the primary outcome, PCBV, as

Neurology: Neuroimmunology & Neuroinflammation 3 estimated by Altmann et al.23 using an SD of 1.51 for SIENA LA cohort were for reasons of claustrophobia during atrophy, 2-year study length with yearly MRIs, and 60% effect MRI, prolonged nausea and vomiting which resolved size. A sample size of 23 per arm was needed to obtain 80% power on cessation of LA, and significant concurrent illness and a significance of p , 0.05. Projected enrollment was (prostate cancer, proteinuria, and worsening renal increased to allow for dropouts. Participants were assigned to LA or placebo in a 1:1 manner by the unblinded research pharmacist function). Because the dropout with claustrophobia following a permutated block randomization based on EDSS did not complete the baseline MRI, the LA cohort #4.5 or .4.5.13 All other study personnel were blinded to the ITT sample size for PCBV was 26 (figure 1). treatment assignment. MRIs were labeled with additional ran- Baseline demographics are presented in table 1. domly generated numbers during analyses to further reduce the The LA and control cohorts were overall matched risk of bias. Statistical analysis was performed by a blinded stat- for age, sex, MS duration, education, and disability. istician for primary, secondary, and safety outcome measures. There were no significant differences between treat- Statistical methods. Intention-to-treat (ITT) analysis used lin- ment arms (all p . 0.05). Table 2 presents the base- ear mixed models to evaluate the effect of LA on annualized line values of the main study outcome measures by PCBV. Mixed models were used to adjust for within- participant serial correlation, to account for the repeated meas- study arms. The LA cohort had a significantly larger urements of the longitudinal design, and to include all study baseline whole-brain (p 5 0.004) and total deep grey participants. Models were corrected for participant age, sex, and matter volumes (p 5 0.042). Controls had larger MS duration with standard model diagnostics to identify overly baseline normalized T2-lesion volumes. influential leverage points. Multiple comparisons were accounted Compliance with study drug was 87% by pill for using the Holm-Sidak correction within the study outcome counts. Two participants took a halved dose of LA domains. Outlier data points were identified through standard diagnostic techniques using combinations of data point leverage, per protocol for the majority of the study, one each individual residuals, and Cook distance to identify overly influen- for gastritis and elevated alkaline phosphatase levels. tial observations and exclude them from baseline and 2-year change Rate of brain atrophy and secondary imaging outcomes. analyses. Data from participants taking a reduced dose of LA (n 5 2) were not handled differently in outcome analyses as their After 2 years, participants taking LA had significantly limited number made subgroup assessment intractable. Mixed- less annualized PCBV (20.21% [SEE 0.14]) than model results are reported as rates of change with variance repre- controls (20.65% [SEE 0.10], p 5 0.002), with sented by the standard errors of the coefficient estimates (SEEs). the beneficial effect size of LA treatment observed Post hoc analysis of the primary outcome measure was conducted to be a 0.44% 6 0.29% improvement in the rate adding the baseline whole-brain volume and baseline T2-lesion of whole-brain atrophy (95% confidence interval volume as covariates to the model. All analyses were performed – using R 3.3.1 with additional utility from the lme4 package.24,25 [CI] 0.157 0.727). This change corresponds to a 68% reduction in the rate of brain atrophy in LA vs RESULTS Of the 54 consented and randomized, 51 placebo (figure 2A, table 3). In post hoc analysis, participants (27 LA and 24 placebo) took at least 1 baseline whole-brain volume was not associated with dose of study drug and were included in the ITT anal- any of the other controlling covariates (age, sex, and yses (figure 1). Forty-six participants completed the disease duration), nor did it affect the differences in study (22 LA and 24 placebo). The 5 dropouts in the PCBV between the study cohorts. Likewise, T2-

Table 1 Baseline demographics and clinical characteristics of study participants by treatment (n 5 51)

LA (n 5 27) Placebo (n 5 24)

No. or SD, %, or IQR No. or SD, %, or IQR median (range) median (range) Age, y 57.9 6.7 59.7 6.0

Women 16 59 15 63

Caucasian 26 96 23 96

MS duration, y 30.9 9.3 29.1 9.9

Education >high school 25 93 21 88

EDSS 5.5 2.5 (3.0–8.0) 6 1.5 (3.0–9.0)

EDSS £4.5 11 40.8 9 37.5

EDSS 5.0–6.5 12 44.4 11 45.8

EDSS ‡6.5 4 14.8 4 16.7

Taking disease-modifying therapya 12 44 11 46

Abbreviations: EDSS 5 Expanded Disability Status Scale; IQR 5 interquartile range; LA 5 lipoic acid. a LA: glatiramer acetate (n 5 6), interferon-b 1a (n 5 4), interferon-b 1b (n 5 2); placebo: glatiramer acetate (n 5 5), interferon-b 1a (n 5 3), interferon-b 1b (n 5 3).

4 Neurology: Neuroimmunology & Neuroinflammation 2-year changes in PCBV for LA (20.45% [SEE Table 2 Baseline differences between study outcome measures (mean, SD) 0.71]) vs controls (21.31% [SEE 1.10], p 5 0.001)

LA Placebo in study completers are shown (figure 2B). Brain Sample sizea segmentation and OCT did not reveal significant LA, placebo No. SD No. SD differences in annualized rates of change between Normalized whole-brain volume, cubic cmb 26, 24 1,451 63 1,395 66 study arms; the increase in T2-lesion volume in the Total deep grey matter volume, cubic cmc 26, 24 806.4 7.9 764.6 6.6 LA cohort was almost but not significantly different

Cortical thickness, mmd 23, 21 2.3 0.1 2.3 0.1 from controls (414.6 [SEE 201.2] vs 33.1 [SEE 5 Normalized T2-lesion volume, cubic cme 20, 24 10.0 7.3 14.9 8.3 134.7], p 0.058, table 3).

C1 cross-sectional area, squared mmf 22, 23 68.4 11.0 68.2 9.9 Clinical outcomes. Among clinical outcomes, the LA

Cervical spinal cord lesion occupancy, 24, 23 8.8 6.4 11.0 6.7 cohort demonstrated an improvement in the cubic cm T25FW that was almost but not significantly differ- Average RNFL thickness, mm 42, 38 eyes 78.2 9.8 76.3 10.8 ent from controls (20.535 [SEE 0.358] vs 0.137 Average GCIPL thickness, mm 38, 36 eyes 68.8 8.9 65.9 8.9 [SEE 0.247], 95% CI 21.37 to 0.03, p 5 0.060).

SDMT, number correct 27, 24 39.7 11.2 39.1 8.9 AEs. AEs are presented in table e-1. Numbers of AEs Timed 25 Foot Walk, s 23, 19 11.7 11.3 11.5 9.6 were similar between LA and controls (80 and 68, respectively, p 5 0.87), with 6 serious AEs (SAEs) Abbreviations: GCIPL 5 retinal ganglion cell plus inner plexiform layer; LA 5 lipoic acid; RNFL 5 retinal nerve fiber layer; SDMT 5 Symbol Digit Modalities Test. each. More GI upset (14 [SD 17%] vs 2 [SD 3%], a Sample sizes reflect removal of outliers and for exclusion reasons specified in Methods. p 5 0.007) and fewer falls (12 [SD 15%] vs 26 [SD b SIENAX. 38%], p 5 0.03) occurred in the LA cohort. One LA c FIRST FSL. d FreeSurfer. participant developed a nonbothersome vesicular e MIPAV, Lesion-TOADS, semiautomated segmentation. rash, similar to a previously reported one, that f MIPAV. resolved within 6 weeks of study completion.11 One relapse occurred in each study cohort, neither affect- lesion volume change over the study period was not ing the subsequent EDSS examination. There was no associated with any of the correcting covariates, new suicidality by C-SSRS, and no differences including baseline T2-lesion volume. Individual between cohorts in new laboratory abnormalities;

Figure 2 Differences in brain atrophy at 2 years between LA and control cohorts

Annualized percent change brain volume (PCBV) between LA and placebo cohorts using intention-to-treat analysis of 51 participants with secondary pro- gressive MS (A). Two-year PCBV from study completers is shown and demonstrates significantly less PCBV in the LA cohort (n 5 22, 20.45% [SEE 0.71]) than controls (n 5 24, 21.31% [SEE 1.10], p 5 0.001, B). LA 5 lipoic acid; SEE 5 standard errors of the coefficient estimate.

Neurology: Neuroimmunology & Neuroinflammation 5 Table 3 Study outcomes

LA Placebo

No. SEE Number SEE p Value

Percent change brain volumea 20.214 0.143 20.653 0.097 0.002

Total deep grey matter volume, cubic mmb 42.0 268.1 2265.2 185.5 0.25

Cortical thickness percent changec 20.137 0.339 20.158 0.24 0.95

T2-lesion volume, cubic cmd 414.6 201.2 33.1 134.7 0.058

C1 cross-sectional area, squared mme 20.113 0.774 20.984 0.52 0.26

Average RNFL, mm 20.279 0.744 20.286 0.518 0.99

Average GCIPL, mm 20.322 0.595 20.145 0.417 0.77

Timed 25-Foot Walk, s 20.535 0.358 0.137 0.247 0.060

ABC, units 2.51 1.679 0.177 1.174 0.16

MSWS-12, units 2.49 1.36 2.99 0.95 0.71

RAND, physical, units 0.589 0.876 20.667 0.61 0.15

RAND, mental, units 20.385 1.187 0.522 0.827 0.45

SDMT, units 2.09 0.93 2.59 0.64 0.59

EDSS stable or improved 61% — 71% — OR 1.33, p 5 0.77

Taking disease-modifying therapyf 11 5% 11 46% —

Abbreviations: ABC 5 Activities of Balance Confidence; EDSS 5 Expanded Disability Status Scale; GCIPL 5 retinal ganglion cell plus inner plexiform layer; LA 5 lipoic acid; MSWS-12 5 Multiple Sclerosis Walking Scale 12; OR 5 odds ratio; RAND 5 RAND 36-Item Short Form Health Survey; RNFL 5 retinal nerve fiber layer; SEE 5 standard errors of the coefficient estimate; SDMT 5 Symbol Digit Modalities Test. Annualized rates of change (SD) for MRI and clinical outcomes taken from intention-to-treat mixed models across the 2- year study period. Significant differences are indicated in italics. Rates of change, SEEs, and p values taken from model corrected outputs. a SIENAX. b FIRST FSL. c FreeSurfer. d MIPAV, Lesion-TOADS, semiautomated segmentation. e MIPAV. f LA cohort: dimethyl fumarate (n 5 1), glatiramer acetate (n 5 5), interferon-b 1a (n 5 3), interferon-b 1b (n 5 2); placebo cohort: glatiramer acetate (n 5 5), interferon-b 1a (n 5 3), interferon-b 1b (n 5 3).

however, only those taking LA required a dose reduc- were fewer falls in the LA cohort. Overall LA was safe, tion per protocol (n 5 2). The SAE thought directly well tolerated, had high compliance, and had no related to LA was vomiting and dehydration requiring unexpected deleterious AEs or SAEs attributed to LA. hospitalization which resolved once stopping LA. The reduction in the brain atrophy rate compares Notable AEs leading to dropout were one LA patient favorably with a large phase 3 trial of ocrelizumab with baseline elevated creatinine progressing to renal (n 5 731), reporting a 17.5% reduction in the failure and another with proteinuria due to glomeru- whole-brain atrophy rate over 120 weeks.26 In contrast lonephritis. A consulting nephrologist did not think to the present study, the ocrelizumab trial included that the 2 AEs were related to LA. AEs are discussed only primary progressive MS patients with more men further in appendix e-2. (51% vs 39%) who were younger (45 vs 59 years), had shorter disease duration (6.5 vs 30 years), and less dis- DISCUSSION This two-year, randomized, double- ability (EDSS 4.7 vs 5.4). The ocrelizumab trial limited blind, placebo-controlled pilot trial demonstrated inclusion to those with inflammatory CSF, a character- a significant reduction in the primary outcome istic not assessed in this study. Other disease-modifying measure of the annualized rate of whole-brain atrophy therapy trials in progressive MS populations have not by SIENA in people with SPMS taking 1,200 mg demonstrated robust changes in brain atrophy rates or daily oral LA. The annualized PCBV reduction was have not used this outcome measure.27–29 not influenced by the planned covariates of age, sex, The PCBV reduction was for the whole-brain and disease duration, nor by the post hoc additions of assessment by SIENA, a robust registration–based baseline brain and T2-lesion volumes. There was longitudinal atrophy measure.30 Brain segmentation a suggestion of improved T25FW time, and there did not reveal specific brain compartments

6 Neurology: Neuroimmunology & Neuroinflammation responsible for differential atrophy rates. One expla- study outcomes. A larger sample size in future studies nation is that the sample size calculation was based on should correct for baseline imbalances. Finally, as SIENA and not on segmentation methods, techni- a contrast agent was not used, baseline and ongoing ques which produce heterogeneous results and differences in inflammatory activity on MRI are require larger samples.30 Segmented volumes may be unknown. more prone to physiologic factors including hydra- This 2-year pilot trial of 1,200 mg daily LA dem- tion status and positioning than whole-brain meas- onstrated significant reduction in PCBV in patients urements.31 Alternatively, the PCBV reduction may with SPMS over controls. The small sample size pre- come from an effect common across all brain tissues. cluded detection of clinical benefits, although there Study replication in a larger sample will clarify the was suggestion of improved walking times and signif- nature of LA’s atrophy effects. icantly fewer falls. Although LA was overall safe and The suggestion of greater increase in T2-lesion well tolerated, future studies need to establish clinical volume in the LA cohort is of uncertain significance. benefits and explore mechanisms of action of LA in The increase may be real and thus represent potential progressive MS. harmful effects of LA, may represent physiologic changes other than due to MS, may result from AUTHOR CONTRIBUTIONS MR postprocessing quality issues, or may not be dif- Rebecca Spain: study concept and design, study supervision, analysis and interpretation of data, drafting of manuscript, and obtained funding. Ka- 15 ferent from placebo. Further evaluation is therine Powers: data acquisition, MRI analyses, and drafting of manu- warranted. script. Charles Murchison: statistical analyses and drafting of the Overall, LA was safe and well tolerated. Labora- manuscript. Elizabeth Heriza: data acquisition and manuscript review. Kimberly Winges: review of OCTs, data analysis and interpretation of tory abnormalities were limited to asymptomatic ele- data, and manuscript review. Vijayshree Yadav, Michelle Cameron, and vations in alkaline phosphatase which improved on Edward Kim: data acquisition. Fay Horak: study concept and manuscript cessation of LA. Although there were no unexpected review. Jack Simon: study concept and design, analysis and interpretation AEs or SAEs clearly attributable to LA, the renal fail- of data, and drafting of manuscript. Dennis Bourdette: study concept and design, study supervision, analysis and interpretation of data, and critical ure and glomerulonephritis cases raise concern. Previ- review of the manuscript. ous studies with LA have not reported renal – dysfunction.9 11 Even so, closer renal function mon- ACKNOWLEDGMENT itoring is prudent for future LA studies and suggests Pure Encapsulations, Sudbury, MA, provided the lipoic acid and placebo. caution in recommendation of this dose of LA before The authors thank the study participants, Dr. Elizabeth Heriza for study management, and the members of the data safety and monitoring board further evaluations. for annual reviews of data. A potential study confounder of which the authors were unaware when planning the study was the pres- STUDY FUNDING ence of quercetin (8.6 mg daily) in the placebo. Quer- Supported by the Department of Veterans Affairs (B7493-W, R. Spain), cetin, like other bioflavonoids, is biologically active. NIH (UL1TR000128). Although conflicting reports exist, Van Beek et al. DISCLOSURE found that 10 mg daily oral quercetin increased R. Spain received research support from Department of Veterans Affairs, inflammation and exacerbated active and passive Oregon Clinical and Translational Research Institute, VA Portland EAE in mice. This raises the possibility that in the Health Care System, Oregon Health & Science University, National present study, quercetin could have worsened the MS Society, Conrad Hilton Foundation, Medical Research Foundation 32 of Oregon, and Race to Erase MS. K. Powers, C. Murchison, E. Heriza, controls and exaggerated the effects of LA. The cal- and K. Winges report no disclosures. V. Yadav consulted for Bayer, culated analogous human dose to the Van Beek study Biogen, and Teva MS; served on the speaker’s bureau for Novartis and is 2 g, orders of magnitude greater than the placebo Biogen; and received research support from Biogen, Department of Vet- capsules.33 Therefore, while unlikely to have affected erans Affairs, McDougall Foundation, National MS Society, Race to Erase MS Foundation, and NIH. M. Cameron received travel funding present study outcomes, future studies should avoid from the Consortium of Multiple Sclerosis Centers and VA Portland quercetin to exclude this possibility. Research; is an editorial board member for the Journal of Hand Therapy; Other study limitations relate primarily to the section editor for Nature Reviews Neuroscience; receives publishing royal- small sample size of this pilot study with resulting loss ties from Elsevier; consulted for ReWalk Corporation and Adamas Cor- poration; received research support from the Department of Veterans of power to detect clinical and secondary imaging Affairs, Rehabilitation Research and Development Service, National outcomes as has been discussed. In light of this, the MS Society, and Paralyzed Veterans of America Foundation. E. Kim very robust reduction in the brain atrophy rate might served on the scientific advisory board for Genzyme and Teva and received research support from the National MS Society. F. Horak served be viewed with skepticism, although the comparable on the scientific advisory board for MS Society and NIH NCMRR; atrophy rate of the controls to other untreated SPMS received travel funding and/or speaker honoraria for lectures and educa- cohorts lends credence to the findings.23,34 Baseline tional activities not funded by industry; served as an associate editor for differences in brain volume or T2-lesion volume ex- Gait and Posture, Cerebellum, and Encyclopedia for Neuroscience; served on the editorial board for Journal of Biomechanics, Frontiers in Neurology, and isted between study cohorts; however, post hoc anal- Frontiers in Neuro-otology; holds patents for device for conditioning bal- yses did not find an influence of these differences on ance and motor coordination, Instrumented mobility system to

Neurology: Neuroimmunology & Neuroinflammation 7 objectively measure balance and gait; receives publishing royalties from 14. Schulz KF, Altman DG, Moher D, Group C. CON- McGraw-Hill; received research support from MRF NCI, NIH, SBIR, SORT 2010 statement: updated guidelines for reporting MRF, MRF Mentor, STTR, NIA, NCI, NIH/CCHD/NCMRR MJFF, parallel group randomized trials. Ann Intern Med 2010; NMSS, National Institute of Neurological Disorders and Stroke, NMSS 152:726–732. Mentor, and US Army MRAA; and holds stock in APDM, Inc. J. Simon 15. Shiee N, Bazin PL, Ozturk A, Reich DS, Calabresi PA, served on the scientific advisory board for Biogen; receives publishing Pham DL. A topology-preserving approach to the segmen- royalties from Cambridge University Press; and consulted for Biogen, Guthy-Jackson Charitable Foundation. D. Bourdette received travel tation of brain images with multiple sclerosis lesions. Neu- – funding from National Multiple Sclerosis Society, Consortium of MS roimage 2010;49:1524 1535. Centers, and Paralyzed Veterans of America; serves on the editorial board 16. Smith SM, Jenkinson M, Woolrich MW, et al. Advances for Neurology; holds patents for treatment of multiple sclerosis with cyclic in functional and structural MR image analysis and im- peptide derivatives of cyclosporin and thyromimetic drugs for stimulating plementation as FSL. Neuroimage 2004;23(suppl 1): remyelination in multiple sclerosis; consulted for Magellan Health, Best S208–S219. Doctors, Inc.; and received research support from National MS Society. 17. Reuter M, Schmansky NJ, Rosas HD, Fischl B. Within- Go to Neurology.org/nn for full disclosure forms. subject template estimation for unbiased longitudinal image analysis. Neuroimage 2012;61:1402–1418. Received March 7, 2017. Accepted in final form May 15, 2017. 18. Fischer JS, Rudick RA, Cutter GR, Reingold SC. The Multiple Sclerosis Functional Composite Measure REFERENCES (MSFC): an integrated approach to MS clinical outcome 1. Mahad DH, Trapp BD, Lassmann H. Pathological mech- assessment. National MS Society Clinical Outcomes anisms in progressive multiple sclerosis. Lancet Neurol Assessment Task Force. Mult Scler 1999;5:244–250. 2015;14:183–193. 19. Powell LE, Myers AM. The Activities-specific Balance 2. Furby J, Hayton T, Anderson V, et al. Magnetic resonance Confidence (ABC) Scale. J Gerontol A Biol Sci Med Sci imaging measures of brain and spinal cord atrophy corre- 1995;50A:M28–M34. late with clinical impairment in secondary progressive mul- 20. Hobart JC, Riazi A, Lamping DL, Fitzpatrick R, Thompson tiple sclerosis. Mult Scler 2008;14:1068–1075. AJ. Measuring the impact of MS on walking ability: 3. Biewenga GP, Haenen GR, Bast A. The pharmacology of the 12-Item MS Walking Scale (MSWS-12). Neurology the antioxidant lipoic acid. Gen Pharmacol 1997;29: 2003;60:31–36. 315–331. 21. Parmenter BA, Weinstock-Guttman B, Garg N, 4. Biewenga GP, Dorstijn MA, Verhagen JV, Haenen GR, Munschauer F, Benedict RH. Screening for cognitive Bast A. Reduction of lipoic acid by lipoamide dehydroge- impairment in multiple sclerosis using the Symbol digit nase. Biochem Pharmacol 1996;51:233–238. Modalities Test. Mult Scler 2007;13:52–57. 5. Schreibelt G, Musters RJ, Reijerkerk A, et al. Lipoic acid 22. RAND Corporation. 36-Item Short Form Survey (SF-36). affects cellular migration into the central nervous system Available at: rand.org/health/surveys_tools/mos/36-item- and stabilizes blood-brain barrier integrity. J Immunol short-form.html. Accessed May 2, 2017. 2006;177:2630–2637. 23. Altmann DR, Jasperse B, Barkhof F, et al. Sample sizes for 6. Shay KP, Michels AJ, Li W, Kong AN, Hagen TM. Cap- brain atrophy outcomes in trials for secondary progressive independent Nrf2 translation is part of a lipoic acid- multiple sclerosis. Neurology 2009;72:595–601. stimulated detoxification stress response. Biochim Biophys 24. R Core Team. R: A Language and Environment for Sta- Acta 2012;1823:1102–1109. tistical Computing. Vienna: R Foundation for Statistical 7. Morini M, Roccatagliata L, Dell’Eva R, et al. Alpha-lipoic Computing [Internet]; 2016. Available at: R-project.org/. acid is effective in prevention and treatment of experimen- Accessed May 2, 2017. tal autoimmune encephalomyelitis. J Neuroimmunol 25. Bates D, Mächler M, Bolker B, Walker S. Fitting linear 2004;148:146–153. mixed-effects models using lme4. J Stat Softw 2015;67:1–48. 8. Marracci GH, Jones RE, McKeon GP, Bourdette DN. 26. Montalban X, Hemmer B, Rammohan K, et al. Efficacy Alpha lipoic acid inhibits T cell migration into the spinal and safety of ocrelizumab in primary progressive multiple cord and suppresses and treats experimental autoimmune sclerosis—results of the placebo-controlled, double-blind, encephalomyelitis. J Neuroimmunol 2002;131:104–114. Phase III ORATORIO study. Barcelona, Spain: EC- 9. Yadav V, Marracci GH, Munar MY, et al. Pharmacoki- TRIMS Online Library; 2015. netic study of lipoic acid in multiple sclerosis: comparing 27. Kapoor R, Furby J, Hayton T, et al. Lamotrigine for mice and human pharmacokinetic parameters. Mult Scler neuroprotection in secondary progressive multiple sclero- 2010;16:387–397. sis: a randomised, double-blind, placebo-controlled, 10. Reljanovic M, Reichel G, Rett K, et al. Treatment of diabetic parallel-group trial. Lancet Neurol 2010;9:681–688. polyneuropathy with the antioxidant thioctic acid (alpha-lipoic 28. Chataway J, Schuerer N, Alsanousi A, et al. Effect of acid): a two year multicenter randomized double-blind high-dose simvastatin on brain atrophy and disability placebo-controlled trial (ALADIN II). Alpha Lipoic Acid in in secondary progressive multiple sclerosis (MS-STAT): Diabetic Neuropathy. Free Radic Res 1999;31:171–179. a randomised, placebo-controlled, phase 2 trial. Lancet 11. Yadav V, Marracci G, Lovera J, et al. Lipoic acid in mul- 2014;383:2213–2221. tiple sclerosis: a pilot study. Mult Scler 2005;11:159–165. 29. Tourbah A, Lebrun-Frenay C, Edan G, et al. MD1003 12. Polman CH, Reingold SC, Edan G, et al. Diagnostic (high-dose biotin) for the treatment of progressive multiple criteria for multiple sclerosis: 2005 revisions to the sclerosis: a randomised, double-blind, placebo-controlled “McDonald Criteria.” Ann Neurol 2005;58:840–846. study. Mult Scler 2016;22:1719–1731. 13. Kurtzke JF. Rating neurologic impairment in multiple 30. Durand-Dubief F, Belaroussi B, Armspach JP, et al. Reli- sclerosis: an expanded disability status scale (EDSS). Neu- ability of longitudinal brain volume loss measurements rology 1983;33:1444–1452. between 2 sites in patients with multiple sclerosis:

8 Neurology: Neuroimmunology & Neuroinflammation comparison of 7 quantification techniques. AJNR Am J encephalomyelitis in SJL mice. Biochem Pharmacol Neuroradiol 2012;33:1918–1924. 2005;70:220–228. 31. Sampat MP, Healy BC, Meier DS, Dell’Oglio E, Liguori 33. Nair AB, Jacob S. A simple practice guide for dose con- M, Guttmann CR. Disease modeling in multiple sclerosis: version between animals and human. J Basic Clin Pharm assessment and quantification of sources of variability in 2016;7:27–31. brain parenchymal fraction measurements. Neuroimage 34. De Stefano N, Giorgio A, Battaglini M, et al. Assessing 2010;52:1367–1373. brain atrophy rates in a large population of untreated 32. Verbeek R, van Tol EA, van Noort JM. Oral flavonoids multiple sclerosis subtypes. Neurology 2010;74:1868– delay recovery from experimental autoimmune 1876.

Neurology: Neuroimmunology & Neuroinflammation 9 Glucocorticoid-associated blood glucose response and MS relapse recovery

Myla D. Goldman, MD, ABSTRACT MSc Objective: To determine the relationship between MS relapse recovery and blood glucose (BG) Scott Koenig, BS response to IV methylprednisolone (IVMP) treatment. Casey Engel, BS Methods: We retrospectively identified 36 patients with MS admitted for IVMP treatment of Christopher R. acute relapse who had adequate data to characterize BG response, relapse severity, and recov- McCartney, MD ery. The relationship between glucocorticoid-associated nonfasting BG (NFBG) and relapse Min-Woong Sohn, PhD recovery was assessed. Results: Highest recorded nonfasting BG (maximum NFBG [maxNFBG]) values were significantly Correspondence to higher in patients with MS without relapse recovery compared with those with recovery (271 6 Dr. Goldman: 68 vs 209 6 48 mg/dL, respectively; p 5 0.0045). After adjusting for relapse severity, MS patients [email protected] with maxNFBG below the group median were 6 times (OR 5 6.01; 95% CI, 1.08–33.40; p 5 0.040) more likely to experience relapse recovery than those with maxNFBG above the group median. In a multiple regression model adjusting for age, sex, and relapse severity, a 1-mg/dL increase in the maxNFBG was associated with 4.5% decrease in the probability of recovery (OR 5 0.955; 95% CI, 0.928–0.983; p 5 0.002). Conclusions: These findings suggest that higher glucocorticoid-associated NFBG values in acutely relapsing patients with MS are associated with diminished probability of recovery. This relationship could reflect steroid-associated hyperglycemia and/or insulin resistance, defects in non–steroid-associated (e.g., prerelapse) glucose metabolism, or both. This study included only those admitted for an MS relapse, and it is this subset of patients for whom these findings may be most relevant. A prospective study to evaluate glucose regulation and MS relapse recovery in a broader outpatient MS population is under way. Neurol Neuroimmunol Neuroinflamm 2017;4:e378; doi: 10.1212/NXI.0000000000000378

GLOSSARY BG 5 blood glucose; EDSS 5 Expanded Disability Status Scale; EMR 5 electronic medical record; FSS 5 Functional System Score; ICD-9 5 International Classification of Diseases–9; IVMP 5 IV methylprednisolone; maxNFBG 5 maximum NFBG; NFBG 5 nonfasting BG.

Comorbid medical conditions can substantially affect neurologic disease outcomes. In most studies addressing the relationship between glucose regulation and patient outcomes in neuro- logic disorders (e.g., stroke), higher blood glucose (BG) concentration at presentation is associ- ated with poorer outcomes.1–8 Although this has not been specifically addressed in MS, diabetes and cardiovascular comorbidities are associated with accelerated MS disease progression and development of disability.9–11 The standard treatment for MS relapse is IV methylprednisolone (IVMP), 1,000 mg/d for 3–5 days. Hyperglycemia is a common consequence of glucocorticoid administration, largely attributed to reductions in insulin sensitivity.12 Individuals with pretreatment insulin resistance

From the Department of Neurology (M.D.G.), Division of Endocrinology and Metabolism, Department of Medicine (C.R.M.), and Department of Public Health Sciences (M.-W.S.), University of Virginia School of Medicine, Charlottesville; University of Maryland School of Medicine (S.K.), Baltimore; and University of Virginia College of Arts and Sciences (C.E.), Charlottesville. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by ziMS Foundation (MDG). This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 are more likely to exhibit a hyperglycemic considered to be an NFBG. Because of our inability to get response, and risk factors for insulin resistance reliable information on the frequency and dosing of sliding scale insulin administration, we chose to focus on the highest measured (e.g., reduced mobility, sedentary lifestyle, and NFBG during glucocorticoid treatment (maximum NFBG repeated glucocorticoid exposure) are com- [maxNFBG]). Given the very high doses and long duration of mon in MS. Despite the potential risk of action of MP, we suspect that time since administration did not hyperglycemia, IVMP is commonly adminis- play a major role in differential glucose responses, at least within the 18-hour time frame of our study. tered in the outpatient setting without glucose An Expanded Disability Status Scale (EDSS)-certified neurol- monitoring.13 To our knowledge, no pub- ogist (MDG), who was masked to BG values, classified patients lished studies have addressed the relationship with MS as recovered (full, partial, or none) retrospectively using outpatient and inpatient medical records and examinations of vis- between (1) glucose regulation in the setting of its occurring before, during, and after the acute relapse. Prerelapse glucocorticoid treatment of acute MS relapse and postrelapse examinations were taken from available records and (2) MS relapse recovery outcomes. obtained within 6 months of relapse assessment. The recovery Despite standardized treatment, the degree classification was based on the compendium of available event- related data collected from clinical notes before, during, and after of recovery from any single MS relapse is both relapse, including patient reports and treating physician impres- variable and unpredictable, but the likelihood sions, neurologic examinations, and motor strength testing in of residual deficits after an MS relapse ranges available physical therapy records focused on the relapse- from 40% to 57%.13–16 Although relapse affected area (e.g., documented return to full power in an affected limb). When the available data indicated a return to prerelapse severity appears to predict recovery, data are function, the patient was classified as recovered. When the data inconsistent regarding the impact of other fac- indicated persistent neurologic deficits, examinations were com- tors, such as sex, age, and site of relapse.13,15,17 pared with prerelapse records to determine whether there was no or partial recovery. To provide further granularity, relapse- We hypothesized that high glucocorticoid- relevant EDSS Functional System Scores (FSS) before, during, associated BG responses would be associated and after the acute relapse were extracted from chart-based neu- with reduced degree of relapse recovery. rologic examinations and documented on the standard neurosta- tus scoring form (neurostatus.net/scoring/index.php). When needed (e.g., in cases of optic neuritis), additional ophthalmology METHODS Standard protocol approvals, registrations, notes for visual acuity were used for the visual FSS. Relapse and patient consents. This retrospective chart review study was severity was then calculated for prerelapse, intrarelapse, and post- approved by the Institutional Review Board at the University of relapse FSS domains (e.g., vision, brainstem, pyramidal, sensory, Virginia, and the study was executed in accordance with the or cerebellar) based on the standard neurostatus scoring. Severity Health Insurance Portability and Accountability Act of 1996. was calculated by subtracting the prerelapse FSS from the intra- We performed an administrative database search (2008– relapse FSS for relapse-relevant domains; similarly, FSS-based 2015) to identify patients admitted to the inpatient service with recovery was calculated by subtracting the postrelapse FSS from a diagnostic code for MS between 2008 and June 2015 (MS the prerelapse FSS for relapse-relevant domains. For patients with International Classification of Diseases–9 [ICD-9] codes: 340, more than one relapse-relevant FSS, an average was calculated for 341, 341.1, 341.2, 341.21, and 341.22). Patient and disease both severity and recovery of the FSS-based scores. Using these characteristics were abstracted from our electronic medical re- FSS obtained from through neurologic examinations available in cords (EMRs) using the following eligibility criteria: age 18–55 the medical records, the EDSS scores were calculated based on years (inclusive), confirmed diagnosis of MS, administration of these 8 FSS in the standard fashion. IVMP 1,000 mg/d diluted in 0.9% normal saline (minimum 1 We compared patient characteristics using 1-way analysis of day exposure), serum BG values drawn prior to glucocorticoid variance, t tests, or x2 tests as appropriate. Logistic regression treatment from a basic or comprehensive metabolic panel, avail- analysis was used to examine the association between the ability of at least 1 nonfasting BG (NFBG) level (i.e., drawn maxNFBG and relapse recovery. We estimated 1 unadjusted between 09:00 and 24:00 hours) obtained within 18 hours after model to predict recovery using the highest NFBG alone and 2 recorded glucocorticoid administration, no known diagnosis of other models that adjusted for relapse severity (mild, intermedi- diabetes, absence of glucose-regulating medications (e.g., metfor- ate, and severe) and demographic variables such as age and sex. min), and availability of demographic data in the EMR (e.g., age We additionally grouped patients into high and low maxNFBG and sex). using the median maxNFBG, thereafter using logistic regression Data collected from the EMR included age, sex, height, to compare the probability of recovery between low and high weight, glucocorticoid dose and number of days treated, time maxNFBG. For all logistic regression models, we computed the of glucocorticoid administration, and if sliding scale insulin was area under the receiver operating characteristic curves as well as administered during hospitalization. Postglucocorticoid BG was the Hosmer-Lemeshow goodness of fit statistic. We used SAS defined as any BG drawn within 18 hours of glucocorticoid 9.2.6 (SAS Institute, Cary, NC) and Stata SE 14 (Statacorp, administration. (The half-life of MP has an estimated range of College Station, TX) for statistical analysis and graphing. 18–24 hours.) Glucocorticoid-associated hyperglycemia is most prominent in the postprandial state, with relatively mild changes RESULTS Our administrative database search identi- in fasting BG; thus, to provide a more sensitive assessment of glucocorticoid-associated BG changes, we classified BG values as fied 413 patients with one of the 6 searched ICD-9 either fasting or nonfasting. Since inpatient breakfast is routinely served codes. Among those, we identified 68 nondiabetic at 8:00 AM, any BG measured between 9:00 AM and 24:00 AM was MS patients who were admitted for MS relapse

2 Neurology: Neuroimmunology & Neuroinflammation Table 1 Demographic characteristics and nonfasting blood glucose levels for the study sample (n 5 36)

Full recovery, n 5 17 Partial recovery, n 5 10 No recovery, n 5 9 p Value

Age, y, mean 6 SD 33.7 6 9.3 35.3 6 10.9 41.3 6 12.4 0.23

Sex, female:male (% female) 14:3 (82.1) 9:1 (90) 4:5 (44.4) 0.05

Body mass index, kg/m2, mean 6 SD 29.8 6 5.6 29.6 6 8.5a 28.2 6 7.8b 0.86

Days treated, mean 6 SD 3.6 6 1.0 3.6 6 1.0 3.9 6 1.2 0.76

MS disease duration, mean 6 SD 5.2 6 6.0 4.6 6 6.9 6.4 6 6.8 0.82

Time from event to recovery 2.7 6 1.4 3.3 6 2.5 2.8 6 3.1 0.78 assessment, mo, mean 6 SD

Relapse-related EDSS

Mean 6 SD 5.1 6 2.2 6.3 6 1.3 6.1 6 2.3 0.31

Median (25th, 75th quartile) 6.0 (3.0, 6.5) 6.5 (6.0, 7.5) 6.5 (6.5, 7.5)

Relapse severity

Mild, n (%) 6 (35.3) 0 (0.0) 1 (11.1) 0.21

Moderate, n (%) 6 (35.3) 4 (44.4) 5 (55.6)

Severe, n (%) 5 (29.4) 5 (55.6) 3 (33.3)

Relapse-related FSS

Vision, n (%); median FSS 2 (11.8); 4.0 2 (20.0); 3.5 1 (11.1); 1.0 0.07

Brainstem, n (%); median FSS 10 (58.8); 2.0 7 (70.0); 2.0 6 (66.7); 2.0 0.69

Pyramidal, n (%); median FSS 16 (94.1); 3.0 10 (100); 3.0 9 (100); 3.0 0.11

Cerebellar, n (%); median FSS 8 (47.1); 2.0 8 (80.0); 3.0 4 (44.4); 3.0 0.60

Sensory, n (%); median FSS 13 (76.5); 2.0 7 (70.0); 3.0 6 (66.7); 3.0 0.04

Abbreviations: EDSS 5 Expanded Disability Status Scale; FSS 5 Functional System Score. All percentages are of column totals, unless otherwise noted. a n 5 9. b n 5 8. between January 2008 and June 2015. Subsequent likely to be men. Among those with a full recovery, detailed chart review resulted in exclusion of 32 pa- we found an overall lower EDSS and a larger pro- tients for the following reasons: dosing of IVMP 500 portion with mild relapse severity. However, median mg/daily (n 5 1), no recovery data available (n 5 5), relapse EDSS scores were similar (6.0–6.5) for all 3 no blood draws within 18 hours of steroid adminis- recovery groups, and the proportion with moderate tration (n 5 6), no available NFBG values (n 5 19), and severe relapse was well distributed across all 3 and body mass index .40 kg/m2 (n 5 1). This re- recovery groups (table 1). sulted in a total of 36 patient admissions meeting all Table 2 shows glucocorticoid-associated BG levels study eligibility criteria (table 1). The most common for each of the 3 recovery groups. Because of the small cause of exclusion was the absence of required NFBG sample sizes, we dichotomized recovery groups by records. Among the 36 patients used for analysis, only either (1) combining the 2 groups with any recovery 2 (5.6%) had received 2 days of steroids; all other (i.e., some recovery [full or partial] vs no recovery) or patients received $3 days of IVMP. maxNFBG val- (2) combining the 2 groups without complete recov- ues were measured on day 1 or 2 in 29 patients ery (i.e., full recovery vs partial or no recovery) and (72%). There was no difference in the timing of computed their BG levels. The maxNFBG was sig- maxNFBG measurement between recovery groups nificantly different among the 3 recovery groups (p 5 (table 2). 0.018). Dichotomized Group comparison finds a sig- Patients with MS were categorized according to nificant difference between any vs no recovery (p 5 recovery status (full, partial, or no recovery) based 0.005) and a positive trend between complete vs on relapse-related FSS and available medical records. incomplete recovery (p 5 0.076), with lower Table 1 provides details about the demographic char- maxNFBG than the no-recovery group (table 2).We acteristics, relapse severity, and relapse-related FSS by note that correctional (“sliding scale”) insulin is part MS recovery subgroups. Those patients with no of the routine order set for such patients in our insti- recovery from their MS relapse were older and more tution, and 21 patients (;58%) received at least 1

Neurology: Neuroimmunology & Neuroinflammation 3 Table 2 Glucocorticoid-associated BG levels

Full recovery, N 5 17 Partial recovery, N 5 10 No recovery, N 5 9 p Value

Preglucocorticoid fasting BG, mg/dL

Mean 6 SD 93.8 6 10a 96.7 6 10 91.9 6 14.2b 0.31

Median (25th, 75th quartile) 80 (84.5, 101.5) 100 (89, 102) 85.5 (74, 97)

Time from steroid administration to maxNFBG measurement, hrs

Mean 6 SD 8.2 6 5.5 10.8 6 4.2 9.3 6 3.3 0.38

Lowest NFBG, mg/dL 0.82

Mean 6 SD 127.8 6 30.8 135.6 6 34.4 129.3 6 28.4

Median (25th, 75th quartile) 120 (109.5, 145) 131 (110, 148) 128 (111, 149)

Lowest NFBG recovery subgroups, mg/dL Any recovery (n 5 27) No recovery (n 5 9) 0.91

Mean 6 SD 130.7 6 31.8 129.3 6 28.4

Complete (n 5 17) Incomplete (n 5 19) 0.64

127.8 6 30.8 132.6 6 31.0

Average NFBG, mg/dL 0.30

Mean 6 SD 167.6 6 38.4 166.1 6 30.3 188.1 6 30.7

Median (25th, 75th quartile) 154.5 (133.2, 193.0) 158.5 (144.0, 180.0) 182.3 (168.9, 208.9)

Average NFBG recovery Any recovery No recovery 0.12 subgroups, mg/dL (n 5 27) (n 5 9)

Mean 6 SD 167.1 6 35.0 188.1 6 30.7

Complete (n 5 17) Incomplete (n 5 19) 0.45

167.6 6 38.4 176.5 6 31.7

maxNFBG, mg/dL 0.018

Mean 6 SD 205.9 6 56.3 213.2 6 30.8 271.1 6 68.0

Median (25th, 75th quartile) 204.0 (171, 248) 206 (186, 235) 248 (222, 308)

maxNFBG recovery subgroups, mg/dL Any recovery (n 5 27) No recovery (n 5 9) 0.005

Mean 6 SD 208.6 6 47.9 271.1 6 68.0

Complete (n 5 17) Incomplete (n 5 19) 0.079

205.8 6 56.4 240.6 6 58.4

Abbreviations: BG 5 blood glucose; maxNFBG 5 maximum NFBG; NFBG 5 nonfasting blood glucose. a n 5 16. b n 5 6.

dose of insulin during their inpatient stay. This was An additional logistic regression model using one of the main reasons we focused our analysis on maxNFBG as a continuous variable suggested that maxNFBG. Six of 9 (67%) in the no-recovery group a 1-unit increase in maxNFBG (i.e., 1 mg/dL) was and 15 of 27 (56%) in the recovery group received associated with a 4.5% lower likelihood (OR 5 insulin; this apparent difference is in keeping with 0.955; 95% CI, 0.928–0.983; p 5 0.002) of any higher maxNFBG values in the no-recovery group. recovery (model 3 in table 4) after adjusting for Table 3 shows 3 logistic regression models that pre- relapse severity, age, and sex. The estimated probabil- dicted any degree of recovery based on the high vs low ity of any recovery from this model is plotted against maxNFBG groups divided by the median maxNFBG. the maxNFBG in figure, which shows a very strong After adjusting for relapse severity (model 2), patients relationship between maxNFBG and recovery. The with lower maxNFBG were 6 times (OR 5 6.01; 95% figure additionally reveals that all patients with CI, 1.08–33.40; p 5 0.040) more likely to achieve maxNFBG ,200 mg/dL achieved some recovery, some degree of relapse recovery compared with those but no patient with maxNFBG .300 mg/dL experi- with higher maxNFBG. In a multiple regression model enced recovery. that additionally adjusted for age and sex (model 3), the maxNFBG still showed a strong and significant DISCUSSION Patients with MS exhibit remark- association with the probability of recovery (OR 5 able variability with respect to MS relapse fre- 13.69; 95% CI, 1.40–134.12; p 5 0.025). quency and severity. While prior relapse severity

4 Neurology: Neuroimmunology & Neuroinflammation Table 3 Odds ratios and 95% CIs for recovery from acute MS relapse (any vs no recovery) as predicted by high vs low NFBG groups

Model 1 (N 5 36) Model 2 (N 5 35) Model 3 (N 5 35)

Variables OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) p Value

Highest NFBGa (vs low)

High 5.091 (0.864–30.001) 0.072 6.012 (1.082–33.403) 0.040 13.686 (1.396–134.123) 0.025

Severity (vs mild [>21])

Intermediate (21.5 to 21) 0.308 (0.027–3.515) 0.343 0.132 (0.004–4.118) 0.248

Severe (23to21.75) 0.634 (0.053–7.543) 0.718 0.274 (0.008–8.856) 0.465

Age, y

Male 0.866 (0.779–0.963) 0.008

Female 0.933 (0.837–1.039) 0.205

Hosmer-Lemeshow x2 b 1.79 (p 5 0.618) 2.92 (p 5 0.939)

C-statistic 0.685 0.748 0.872

Abbreviations: BG 5 blood glucose; NFBG 5 nonfasting BG. a Highest NFBG with high and low groups was defined using the median. b There are only 2 distinct quantiles because of ties and the x2 statistic could not be computed.

is often associated with future relapse severity,17 this study used a small retrospective sample, the rela- risk factors for relapse severity are not fully under- tionship between BG and relapse recovery is clear and stood. Similarly, the degree of recovery from any significant. Higher glucocorticoid-associated BG lev- single MS relapse is both variable and unpredict- els were associated with a substantially lower proba- able. It is important that the likelihood of residual bility of recovery. Indeed, every patient with a peak deficits after an MS relapse ranges from 40% to NFBG ,200 mg/dL recovered, while none with 57%.13–16 Poor recovery with residual neurologic def- .300 mg/dL did. icits from any single MS relapse contributes to cumu- There is increased recognition that comorbidities lative disability and portends a poorer long-term are significant factors in the MS disease course and prognosis.14,15,18–20 Thus, understanding the deter- associated with worsened outcomes.9–11 In addition, minants of MS relapse recovery is of critical impor- accumulating data suggest that patients with MS may tance: the identification of modifiable risk factors may have underlying glucose intolerance,2–4,20,21 implying lead to improved MS relapse recovery and reduced that these patients are more insulin resistant and at disability accumulation. higher risk for glucocorticoid-induced hyperglycemia. To the best of our knowledge, this study is the first Our data provide important insights into 1 possible to examine the relationship between glucocorticoid- mechanism that impaired glucose tolerance may be associated BG values and MS relapse recovery. While contributing to increased progression among patients

Table 4 Odds ratios and 95% CIs for recovery from acute MS relapse (any vs no recovery) as predicted by maxNFBG (as a continuous variable)

Model 1 (N 5 36) Model 2 (N 5 35) Model 3 (N 5 35)

Variables OR (95% CI) p Value OR (95% CI) p Value OR (95% CI) p Value

Highest NFBG, mg/dL 0.979 (0.967–0.991) 0.001 0.976 (0.962–0.989) 0.001 0.955 (0.928–0.983) 0.002

Severity (vs mild [>21])

Intermediate (21.5 to 21) 0.590 (0.046–7.603) 0.685 0.466 (0.013–17.301) 0.679

Severe (23to21.75) 2.449 (0.127–47.182) 0.553 5.558 (0.134–231.266) 0.367

Age, y

Male 0.843 (0.740–0.961) 0.011

Female 0.924 (0.814–1.049) 0.223

Hosmer-Lemeshow x2 4.50 (p 5 0.810) 4.05 (p 5 0.852) 11.32 (p 5 0.184)

C-statistic 0.780 0.821 0.927

Abbreviations: FSS 5 Functional System Score; NFBG 5 nonfasting blood glucose; Severity 5 mean prerelapse FSS 2 mean relapse FSS.

Neurology: Neuroimmunology & Neuroinflammation 5 not appear to drive the observed association between Figure Scatter plot of predicted probability of recovery from an acute MS relapse and maximum nonfasting blood glucose level (n 5 36) maxNFBG and recovery. In addition, given the very high dose of MP used and its very rapid effects of glu- cose tolerance, we do not anticipate that the length of time on MP (or cumulative dose) affected the BG values. The retrospective nature of our study also limited the number of patients with MS available for inclu- sion, primarily because of 2 issues: (1) most patients with MS in our practice receive glucocorticoids in the outpatient setting and (2) our routine inpatient practice is to draw laboratory test results in the early morning, limiting the numbers of patients who had postprandial BG values available. These limitations resulted in a more severe MS relapsing cohort, and it is this subset of the MS relapsing population for whom these findings may be most relevant. Future work will need to confirm and validate these find- ings in patients with MS receiving outpatient MS Probability of recovery was estimated from model 3, shown in table 3. relapse management. Finally, FSS and EDSS assess- ments were performed retrospectively based on with MS, even those without a confirmed diagnosis chart reviews and may lack accuracy a prospective of diabetes. Although our study strongly suggests assessment can provide, but the assessor was masked a meaningful relationship between glucose regula- to the patient BG levels to mitigate against poten- tion and relapse outcomes, our retrospective data do tial bias. In addition, these measures are based on not allow us to assess the relative roles of hyperglyce- and calculated from the completion of a thorough mia vs underlying insulin resistance in relapse recov- neurologic examination. Therefore, we believe it ery, nor do they allow us to assess the relative roles of to be a reasonable strategy to retrospectively calcu- pretreatment abnormalities vs glucocorticoid-related late FSS and EDSS using thorough examinations (on-treatment) abnormalities. Prospective studies are that were performed and available in the medical needed to further characterize relative contributions records. of these risk factors to MS relapse recovery and, sub- This study represents the first to demonstrate a sig- sequently, the potential benefit of BG regulation dur- nificant association between glucocorticoid-associated ing acute MS treatment. Based on the findings BG responses and the likelihood of relapse recovery in described herein, this appears to be a worthy area of patients with MS. Future prospective studies are research. Indeed, we suggest that this work represents under way to confirm and extend these findings. the first look into a potentially modifiable factor, treatment of which could improve recovery from an AUTHOR CONTRIBUTIONS individual relapse, which might ultimately mitigate the M.D. Goldman: study concept development and design, statistical anal- accumulation of relapse-related disability. Our findings ysis and interpretation, and manuscript drafting and revising. S. Koenig: data collection and manuscript preparation and revision. C. Engle: data also raise concerns about the current clinical practice collection and manuscript preparation and revision. C.R. McCartney: study under which increasingly more patients are being trea- results interpretation and critical revision of manuscript. M.-W. Sohn: data ted with IVMP in outpatient or home-based settings analysis and interpretation and critical revision of manuscript. without BG monitoring. In particular, as the prac- tice of home treatment, routinely without BG or STUDY FUNDING Study was funded in part by the NIH–NINDS (K23NS062898). other monitoring, has been encouraged by an ex- Additional project funding was provided by the ziMS Foundation. panding number of articles that emphasize the sub- 21–24 stantial cost savings. DISCLOSURE The limitations of our study reflect its retrospec- M.D. Goldman served on the scientific advisory board for Novartis, Bio- tive nature. We did not have a reliable source of data gen, Acorda, Questcor, Genzyme, and EMD Serono; received travel for race, a potentially important clinical factor. In funding from Acorda and Biogen; received institutional contracts from Biogen and Novartis; and received research support from NINDS addition, restricting data collection to inpatients with and ziMS Foundation. S. Koenig and C. Engel report no disclosures. clinical laboratory test results potentially introduces C.R. McCartney received research support from NIH.NICHD and a selection bias in our population: those in whom National Multiple Sclerosis Foundation. M.-W. Sohn reports no disclosures. Go to Neurology.org/nn for full disclosure forms. BG was measured may have been most likely to expe- rience hyperglycemia. However, relapse severity did Received November 17, 2016. Accepted in final form May 3, 2017.

6 Neurology: Neuroimmunology & Neuroinflammation REFERENCES American Research Committee on Multiple Sclerosis 1. Selwyn R, Hockenbury N, Jaiswal S, Mathur S, (NARCOMS) Registry. BMC Neurol 2013;13:1–10. Armstrong R, Byrnes K. Mild traumatic brain injury 14. Lublin FB, Baier M, Cutter G. Effect of relapses on devel- results in depressed cerebral glucose uptake: an FDG opment of residual deficit in multiple Sclerosis. Neurology PET study. J Neurotrauma 2013;30:1943–1953. 2003;61:1528–1532. 2. Johnston K, Hall C, Kissela B, Bleck T, Conaway M; 15. Hirst C, Ingram G, Pearson O, Pickersgill T, Scolding N, GRASP Investigators. Glucose regulation in acute stroke Robertson N. Contribution of relapses to disability in patients (GRASP) trial: a randomized pilot trial. Stroke multiple sclerosis. J Neurol 2008;255:280–287. 2009;40:3804–3809. 16. Sellebjerg F, Frederiksen J, Nielsen P, Olesen J. Double- 3. Baker L, Juneja R, Bruno A. Management of hyperglyce- blind, randomized, placebo-controlled study of oral, high- mia in acute ischemic stroke. Curr Treat Options Neurol dose methylprednisolone in attacks of MS. Neurology 2011;13:616–628. 1998;51:529–534. 4. Godoy D, Di Napoli M, Rabinstein A. Treating hypergly- 17. Mowry E, Pesic M, Grimes B, Seen S, Bacchetti P, cemia in neurocritical patients: benefits and perils. Neuro- Waubant E. Demyelinating events in early multiple crit Care 2010;13:425–438. sclerosis have inherent severity and recovery. Neurology 5. Sala F, Menna G, Bricolo A, Young W. Role of glycemia in 2009;72:602–608. acute spinal cord injury. Data from a rat experimental model 18. Eriksson M, Andersen O, Runmarker B. Long-term follow and clinical experience. Ann NY Acad Sci 1999;890:133–154. up of patients with clinically isolated syndromes, relapsing- 6. Zhao Q, Zhang X, Wang L. Mild hypothermia therapy remitting and secondary progressive multiple sclerosis. reduces blood glucose and lactate and improves neurologic Mult Scler 2003;9:260–274. outcomes in patients with severe traumatic brain injury. 19. Langer-Gould A, Popat R, Huang S, et al. Clinical and J Crit Care 2011;26:311–315. demographic predictors of long-term disability in patients 7. Johnston K, Connors A Jr, Wagner D, Knaus W, Wang X, with relapsing-remitting multiple sclerosis. Arch Neurol Haley E Jr. A predictive risk model for outcomes of ische- 2006;63:1686–1691. mic stroke. Stroke 2000;31:448–455. 20. Novotna M, Paz Soldán M, Abou Zeid N, et al. Poor early 8. Hacke W, Donnan G, Fieschi C, et al. Association of relapse recovery affects onset of progressive disease course outcome with early stroke treatment: pooled analysis of in multiple sclerosis. Neurology 2015;85:1355. ATLANTIS, ECASS, and NINDS rt-PA stroke trials. 21. O’Brien J, Ward A, Patrick A, Caro J. Cost of managing Lancet 2004;363:768–774. an episode of relapse in multiple sclerosis in the United 9. Marrie R, Horwitz R, Cutter G, Tyry T, Campagnolo D, States. BMC Health Serv Res 2003;3:1–12. Vollmer T. Comorbidity delays diagnosis and increases 22. O’Connell K, Kelly S, Fogarty E, et al. Economic costs disability at diagnosis in MS. Neurology 2009;2:117–124. associated with an MS relapse. Mult Scler Relat Disord 10. Marrie R, Rudick R, Horwitz R, et al. Vascular comorbid- 2014;3:678–683. ity is associated with more rapid disability progression in 23. Chataway J, Porter B, Riazi A, et al. Home versus out- multiple sclerosis. Neurology 2010;13:1041–1047. patient administration of intravenous steroids for multiple- 11. Marrie R, Horwitz R, Cutter G, Tyry T. Cumulative sclerosis relapses: a randomised controlled trial. Lancet impact of comorbidity on quality of life in MS. Acta Neu- Neurol 2006;5:565–571. rol Scand 2012;125:180–186. 24. Creange A, Debouverie M, Jaillon-Riviere V, et al. 12. Kwon S, Hermayer K. Glucocorticoid-Induced Hypergly- Home administration of intravenous methylpredniso- cemia. Am J Med Sci 2013;345:274–277. lone for multiple sclerosis relapses: the experience of 13. Nickerson M, Marrie R. The multiple sclerosis relapse French multiple sclerosis networks. Mult Scler 2009; experience: patient reported outcomes from the North 15:1085–1091.

Neurology: Neuroimmunology & Neuroinflammation 7 Predicting long-term disability outcomes in patients with MS treated with teriflunomide in TEMSO

Maria Pia Sormani, PhD ABSTRACT Philippe Truffinet, MD Objective: To predict long-term disability outcomes in TEMSO core (NCT00134563) and exten- Karthinathan Thangavelu, sion (NCT00803049) studies in patients with relapsing forms of MS treated with teriflunomide. PhD Methods: A post hoc analysis was conducted in a subgroup of patients who received teriflunomide Pascal Rufi, MD in the core study, had MRI and clinical relapse assessments at months 12 (n 5 552) and 18, and Catherine Simonson, entered the extension. Patients were allocated risk scores for disability worsening (DW) after 1 PhD year of teriflunomide treatment: 0 5 low risk; 1 5 intermediate risk; and 2–3 5 high risk, based on Nicola De Stefano, MD, the occurrence of relapses (0 to $2) and/or active (new and enlarging) T -weighted (T w) lesions PhD 2 2 (#3or.3) after the 1-year MRI. Patients in the intermediate-risk group were reclassified as

responders or nonresponders (low or high risk) according to relapses and T2w lesions on the

Correspondence to 18-month MRI. Long-term risk (7 years) of DW was assessed by Kaplan-Meier survival curves. Dr. Sormani: Results: In patients with a score of 2–3, the risk of 12-week–confirmed DW over 7 years was [email protected] significantly higher vs those with a score of 0 (hazard ratio [HR] 5 1.96, p 5 0.0044). Patients reclassified as high risk at month 18 (18.6%) had a significantly higher risk of DW vs those in the low-risk group (81.4%; HR 5 1.92; p 5 0.0004). Conclusions: Over 80% of patients receiving teriflunomide were classified as low risk (respond- ers) and had a significantly lower risk of DW than those at increased risk (nonresponders) over 7

years of follow-up in TEMSO. Close monitoring of relapses and active T2w lesions after short- term teriflunomide treatment predicts a differential rate of subsequent DW long term. ClinicalTrials.gov identifier: TEMSO, NCT00134563; TEMSO extension, NCT00803049. Neurol Neuroimmunol Neuroinflamm 2017;4:e379; doi: 10.1212/NXI.0000000000000379

GLOSSARY

DW 5 disability worsening; EDSS 5 Expanded Disability Status Scale; HR 5 hazard ratio; IFNb 5 interferon beta; T2w 5 T2 weighted.

Reliable predictors of treatment response would facilitate earlier identification of patients with MS that is unresponsive to their existing therapy.1–3 Various clinical and MRI parameters have been used to classify patients according to their response to treatment. The prognostic value of the Rio score, based on outcomes after 1 year of therapy, was demonstrated in patients treated with interferon beta (IFNb)4 or glatiramer acetate.5 A simplified version, the modified Rio score, used – . ascore(0 3) based on the number of new T2-weighted (T2w) lesions (#5or 5) and clinical relapses (0, 1, or 2) occurring after 1 year of therapy.6 Patients treated with IFNb who had higher modified Rio scores after 1 year showed increased risk of disability worsening (DW) over 4–5 years.6,7 A recent analysis of patients treated with IFNb, performed within the MAGNIMS network, re-evaluated the modified Rio score using a large, multicenter, real-world data set from more than 1,200 patients8 and redefined the cutoffs for minimal MRI activity as $3 new T2w lesions on the MRI performed after 1 year from treatment start. The present study

From the Biostatistics Unit (M.P.S.), University of Genoa, Italy; Sanofi Genzyme (P.T., P.R.), Chilly-Mazarin, France; Sanofi Genzyme (K.T.), Cambridge, MA; Fishawack Communications Ltd (C.S.), Abingdon, Oxfordshire, UK; and Department of Medicine, Surgery, and Neurosciences (N.D.S.), University of Siena, Italy. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by the Genzyme. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 evaluated whether this revised scoring system, Standard protocol approvals, registrations, and patient which was developed for IFNb and evaluated consents. TEMSO was conducted in accordance with the Inter- 8 national Conference on Harmonisation Guidelines for Good over the short term, was able to predict dis- Clinical Practice and the Declaration of Helsinki. The protocol ability outcomes for up to 7 years in patients was approved by central and local ethics committees. Patients treated with teriflunomide, a once-daily oral provided written informed consent before the commencement immunomodulator approved for the treat- of TEMSO and, again, before entry into the extension. Clinical Trial (clinicaltrials.gov) identifiers: TEMSO, NCT00134563; ment of relapsing-remitting MS (AUBAGIO; TEMSO extension, NCT00803049. SanofiGenzyme,Cambridge,MA),using data from the TEMSO extension study Defining scoring categories. The MAGNIMS score is a recently published scoring system8 that classified patients after (NCT00803049). 1 year of treatment with IFNb for the risk of disease progression according to the occurrence of relapse (0 to $2) and new and METHODS , Study population and design. TEMSO was enlarging T2w lesions ( 3or$3) on 12-month MRIs. In the a multinational, multicenter, randomized, placebo-controlled, TEMSO study, MRIs were performed more frequently (every 6

double-blind, parallel-group phase 3 study in patients with months), which allowed us to increase the T2w-lesion cutoff to relapsing forms of MS.9 In the core study (NCT00134563), #3or.3, according to the observation that 6-monthly scanning 1,088 patients were randomized 1:1:1 to receive once-daily oral would detect ;30% more lesions than 12-monthly scanning.7 5 5 placebo (n 363), or teriflunomide 7 mg (n 366), or 14 mg The scoring assessment is reported in figure 1. Patients with 5 (n 359), for 108 weeks. a score of 0, 1, or 2–3 were categorized as having a low, inter- Key inclusion criteria allowed for enrollment of patients aged mediate, or high risk of poor treatment response, respectively. – 10 18 55 years who met the McDonald 2001 criteria for relapsing- Patients with a score of 0 were classified as “treatment res- remitting MS with or without progression, had an Expanded ponders,” whereas patients with scores of 2 or 3 were classified as Disability Status Scale (EDSS) score of #5.5, and had $2 clinical “nonresponders.” In addition, patients initially considered to be relapses in the previous 2 years or 1 relapse during the preceding of intermediate risk of disease progression (score 1) were re- year. EDSS scores were determined at screening, baseline visit, classified 6 months later (month 18 from treatment start) as and every 12 weeks thereafter, and at unscheduled visits when responders or nonresponders, based on the algorithm originally patients returned to the clinic for the assessment of potential used to monitor response to IFNb.1 relapse. MRIs were taken at baseline and at weeks 24, 48, 72, In addition, the long-term prognostic value of the use of the and 108. The primary end point was the annualized relapse rate, composite end point of patients with no relapse and no MRI activ- – and the key secondary end point was 12-week confirmed DW. ity (based on active T2w lesions), a concept similar to NEDA (No Patients completing the core study were eligible to enter the Evidence of Disease Activity), during the first year of treatment long-term extension; those already receiving teriflunomide re- with teriflunomide was compared with that of patients with no re-

mained on their original dose, whereas those previously receiving lapses but minimal MRI activity (i.e., 1, 2, or 3 active T2w lesions). placebo were randomized 1:1 to teriflunomide 7 or 14 mg and were treated for up to 9 years.11 Statistical analysis. Based on Kaplan-Meier survival curves, the Patients were included in this post hoc analysis if they risk of DW confirmed for 12 weeks was assessed in patients with received either dose of teriflunomide in the core study, had valid scores that were classified by a score category or as MRI assessments at 1 year, could be assessed for clinical relapses responders/nonresponders. Between-group comparisons were made at year 1, and entered the extension. Data are presented for the using a log-rank test with EDSS strata at baseline and region as combined teriflunomide 7- and 14-mg groups (n 5 552). stratification variables, and risk reductions were estimated from

Figure 1 Scoring assessment

T2w 5 T2-weighted.

2 Neurology: Neuroimmunology & Neuroinflammation of the study, followed by a separation of the curves Figure 2 Patient disposition around 3 years. Over a period of 7 years of follow-up in the TEMSO extension, the probability of 12-week– confirmed DW was significantly increased in the high- risk categories (scores 2 or 3) compared with those in the low-risk categories (score 0) (hazard ratio [HR], 1.96; p 5 0.0044), as well as in the high-risk categories compared with the combined low-risk/intermediate- risk score category (HR, 1.68; p 5 0.0157). In addition, when patients with an intermediate score of 1 were recategorized after an additional 6 months of follow-up, those classified as nonrespond- ers had a significantly higher rate of DW over 7 years compared with those classified as responders (HR, 1.92; p 5 0.0004; figure 3B). Outcomes remained consistent for the group of pa- tients treated with teriflunomide 14 mg only, although because of the small sample size, the differences between groups were not significant (data not shown). a proportional hazards model. Figures with Kaplan-Meier curves The distribution of patients by scoring category at are presented until the end of year 7, after which a few patients (,10%) remained at risk of progression within each group the end of year 1 is shown in figure 4A. Most patients ; presented. ( 90%) were categorized as having low or interme- diate risk of disease progression. Patients with an ini- RESULTS Of the 723 patients treated with terifluno- tial score of 1 (intermediate risk) after 1 year of mide 7 or 14 mg in the core TEMSO study, 551 treatment included those with no relapses and .3

(76.2%) had a computable score at year 1. Of these, active T2w lesions or those with 1 relapse and #3 469 patients were treated in the extension and 299 active T2w lesions. These patients were reclassified 6 completed 7 years of follow-up (figure 2). months later as responders (no relapses and ,2 new

Baseline demographics and clinical disease charac- T2w lesions) or nonresponders ($1 relapse or $2 teristics are shown in table. Patient characteristics new T2w lesions). The majority of these patients were consistent with those of the teriflunomide- (;80%) were reclassified as responders (figure 4B). treated patients in the overall TEMSO population. The responder vs nonresponder classification had Kaplan-Meier estimates of the proportion of pa- a positive predictive value of 67% and a negative pre- tients with DW at the end of year 7, categorized by dictive value of 45%, a sensitivity of 84%, a specificity scoring classification, are shown in figure 3A. The of 24%, and a global accuracy of 63%. probability of disability progression was similar in Finally, because the absence of any disease activity, the score 0 and score 1 categories over the initial years the so-called No Evidence of Disease Activity status,

Table Baseline demographics and disease characteristics

Teriflunomide-treated Teriflunomide-treated patients with computable patients in the overall Characteristic risk score (n 5 551) TEMSO population (n 5 723)

Age, y, mean (SD) 37.7 (8.6) 37.6 (8.6)

Female, n (%) 386 (70.1) 508 (70.8)

No. of relapses 1 year before randomization, mean (SD) 1.3 (0.7) 1.3 (0.7)

Years since first symptoms of MS, mean (SD) 8.78 (6.87) 8.74 (6.78)

MS subtype, n (%)

Relapsing-remitting 511 (92.7) 664 (91.8)

Secondary progressive 20 (3.6) 29 (4.0)

Progressive relapsing 20 (3.6) 30 (4.1)

EDSS score, mean (SD) 2.60 (1.27) 2.68 (1.29)

EDSS score, median (min:max) 2.5 (0.0:5.5) 2.5 (0.0:6.0)

Abbreviation: EDSS 5 Expanded Disability Status Scale.

Neurology: Neuroimmunology & Neuroinflammation 3 Figure 3 Probability of disability worsening by scoring classification

(A) At the end of year 1 and (B) on reclassification of patients with intermediate score 1, 6 months later. *Nonresponders vs responders. HR 5 hazard ratio.

is the emerging target of newer therapies,12,13 we acetate,5,14 and fingolimod,17 clinical relapses and dis- examined whether the proposed cutoff of 3 active ease activity on MRI during the early stages of treat-

T2w lesions might be too conservative for defining ment were predictive of a subsequent differential rate responders among teriflunomide-treated patients. of DW. A small number of active T2w lesions (#3) in This analysis showed no significant difference the absence of relapses had less predictive value, sug- between patients without any relapse or MRI activity, gesting that a small amount of MRI activity can be and those with no relapses and minimal MRI activity tolerated, and a minimal amount of MRI activity

(i.e., 1, 2, or 3 active T2w lesions), with respect to the alone should not be a criterion for switching risk of experiencing DW over the subsequent 7 years treatments. (figure 5). Teriflunomide-treated patients from the TEMSO study who were characterized as having low risk of DISCUSSION Reliable predictors of treatment out- disease progression after 1 year of treatment demon- comes in MS would help to identify patients who strated significantly lower probability of DW over 7 may continue to have disease activity while receiving years of follow-up (median follow-up time was 292 treatment and who may therefore benefit from weeks), when compared with patients who were clas- a change in therapy. Earlier stage treatment decisions, sified as having high risk. Approximately, 26% of pa- when alternative agents are expected to be more effec- tients were classified as having intermediate risk after tive, are likely to improve longer-term outcomes in 1 year; these patients may be more difficult to assess these patients. in terms of response to therapy and subsequent treat- This is the first report of predictive scoring analysis ment choices. An additional evaluation of relapses applied to patients with relapsing forms of MS treated and disease activity on MRI, as applied in previous with teriflunomide. Consistent with earlier reports analyses at 1.5 years,7,18 allowed these patients to be of patients treated with IFNb,4,14–16 glatiramer reclassified as responders or nonresponders, with

Figure 4 Distribution of patients by scoring classification

(A) At the end of year 1 and (B) on reclassification of patients with score 1, 6 months later.

4 Neurology: Neuroimmunology & Neuroinflammation STUDY FUNDING Figure 5 Effect of MRI lesions on the probability of disability worsening in Editorial support for this manuscript was provided by Catherine Simon- patients with no relapses son, of Fishawack Communications Ltd., and was funded by Sanofi Genzyme.

DISCLOSURE M.P. Sormani received travel funding and/or speaker honoraria from Merck Serono, Teva, Genzyme, Novartis, Biogen, and Roche; consulted for Merck Serono, Biogen, Teva, Genzyme, Roche, Novartis, GeNeuro, and MedDay; and received research support from Biogen, TEVA, and Merck Serono. P. Truffinet is employed; receives a salary from Genzyme/Sanofi; and holds stocks in Genzyme/Sanofi. K. Thangavelu is employed by Genzyme Corp and holds stocks in Sanofi. P. Rufi is em- ployed by Sanofi. C. Simonson is employed by Fishawack Communica- tions Ltd. N. De Stefano serves on the scientific advisory board for Merck Serono, Teva, Sanofi, Genzyme, Roche, Biogen, and Novartis Pharma AG; received travel funding from Biogen Idec, Merck Serono S.A., Bayer-Schering AG, Teva, Sanofi, Genzyme, Roche, Biogen, and Novartis Pharma AG; is an editorial advisory board member for Neurological Scien- ces; and received honoraria from Biogen Idec, Merck Serono S.A., Bayer- Schering AG, Teva, Sanofi, Genzyme, Roche, Biogen, and Novartis Pharma AG. Go to Neurology.org/nn for full disclosure forms. HR 5 hazard ratio; T2w 5 T2-weighted. Received January 6, 2017. Accepted in final form May 30, 2017. a greater proportion of TEMSO patients (.81%) REFERENCES being classified as responders. 1. Sormani MP, De Stefano N. Defining and scoring In our analysis, the predictive value of the scoring response to IFN-beta in multiple sclerosis. Nat Rev Neurol classification, particularly in the intermediate group, 2013;9:504–512. extended to over 7 years in the TEMSO long-term 2. Ziemssen T, De Stefano N, Pia Sormani M, Van extension, which extends the finding of studies based Wijmeersch B, Wiendl H, Kieseier BC. Optimizing ther- on a shorter follow-up. A limitation of this analysis is apy early in multiple sclerosis: an evidence-based view. – the loss of patients to follow-up in the extension, Mult Scler Relat Disord 2015;4:460 469. 3. Rio J, Comabella M, Montalban X. Predicting responders although our results support the use of teriflunomide to therapies for multiple sclerosis. Nat Rev Neurol 2009;5: for the long-term treatment of most patients with 553–560. relapsing forms of MS. It could also be suggested that 4. Rio J, Castillo J, Rovira A, et al. Measures in the first year MRI outcomes of patients at 1 year might be better of therapy predict the response to interferon beta in MS. compared with outcomes after steady-state levels of Mult Scler 2009;15:848–853. teriflunomide were achieved (for example, after 6 5.RioJ,RoviraA,TintoreM,etal.Evaluatingthe response to glatiramer acetate in relapsing-remitting months), rather than at baseline. However, to allow multiple sclerosis (RRMS) patients. Mult Scler 2014; for comparisons with other analyses, we followed an 20:1602–1608. 4,7,18 approach consistent with validated methodology. 6. Romeo M, Martinelli V, Rodegher M, et al. Validation of 1- This may also better represent the real-world setting, year predictive score of long-term response to interferon-beta where MRI assessments may be performed less fre- in everyday clinical practice multiple sclerosis patients. Eur J – quently than in clinical trials. Approximately 90% of Neurol 2015;22:973 980. 7. Sormani MP, Rio J, Tintore M, et al. Scoring treatment patients receiving teriflunomide in the TEMSO study response in patients with relapsing multiple sclerosis. Mult were characterized as having low or intermediate risk of Scler 2013;19:605–612. subsequent disease progression. Long-term observation 8. Sormani MP, Gasperini C, Romeo M, et al. Assessing is important in studies of response to treatment and is response to interferon-beta in a multicenter dataset of pa- of particular significance in patients with MS who dem- tients with MS. Neurology 2016;87:134–140. ’ onstrate heterogeneous responses to therapy over time. 9. O Connor P, Wolinsky JS, Confavreux C, et al. Random- ized trial of oral teriflunomide for relapsing multiple scle- rosis. N Engl J Med 2011;365:1293–1303. AUTHOR CONTRIBUTIONS 10. McDonald WI, Compston A, Edan G, et al. Recommen- Design or conceptualization of the study and/or analysis or interpretation of ded diagnostic criteria for multiple sclerosis: guidelines the data: M.P. Sormani, P. Truffinet, K. Thangavelu, P. Rufi, and N. De from the International Panel on the diagnosis of multiple Stefano. Analysis or interpretation of the data: M.P. Sormani, K. Thanga- – velu, and N. De Stefano. Drafting or revising of the manuscript for intel- sclerosis. Ann Neurol 2001;50:121 127. ’ lectual content: M.P. Sormani, P. Truffinet, K. Thangavelu, P. Rufi, and 11. O ConnorP,ComiG,FreedmanMS,etal.Long-term N. De Stefano. Editorial and writing support: C. Simonson. safety and efficacy of teriflunomide: nine-year follow-up of the randomized TEMSO study. Neurology 2016;86: ACKNOWLEDGMENT 920–930. This manuscript was reviewed by Larisa Miller, PharmD, of Sanofi 12. Giovannoni G, Turner B, Gnanapavan S, Offiah C, Genzyme. Schmierer K, Marta M. Is it time to target no evident

Neurology: Neuroimmunology & Neuroinflammation 5 disease activity (NEDA) in multiple sclerosis? Mult Scler interferon beta in multiple sclerosis. Eur J Neurol Relat Disord 2015;4:329–333. 2009;16:1202–1209. 13. Havrdova E, Galetta S, Stefoski D, Comi G. Freedom 16. Sormani MP, Li DK, Bruzzi P, et al. Combined MRI from disease activity in multiple sclerosis. Neurology lesions and relapses as a surrogate for disability in multiple 2010;74(suppl 3):S3–S7. sclerosis. Neurology 2011;77:1684–1690. 14. Romeo M, Martinelli-Boneschi F, Rodegher M, et al. 17. LattanziS,DanniM,CerquaR,TaffiR,ProvincialiL, Clinical and MRI predictors of response to interferon- Silvestrini M. Prediction of disability progression in beta and glatiramer acetate in relapsing-remitting multiple fingolimod-treated patients. J Neurol Sci 2015;358:432–434. sclerosis patients. Eur J Neurol 2013;20:1060–1067. 18. Sormani M, Signori A, Stromillo M, De Stefano N. Refin- 15. Prosperini L, Gallo V, Petsas N, Borriello G, Pozzilli C. ing response to treatment as defined by the modified Rio One-year MRI scan predicts clinical response to score. Mult Scler 2013;19:1246–1247.

6 Neurology: Neuroimmunology & Neuroinflammation High-dose cyclophosphamide without stem cell rescue in immune-mediated necrotizing myopathies

Christopher A. Mecoli, ABSTRACT MD, MHS Objective: To describe the experience managing treatment-refractory immune-mediated necrotiz- Arash H. Lahouti, MD ing myopathies (IMNM) with high-dose cyclophosphamide (HiCy) therapy. Robert A. Brodsky, MD Methods: Five patients with severe refractory IMNM who were treated with HiCy without stem Andrew L. Mammen, cell rescue were identified. Their medical records were reviewed to assess demographic, clinical, MD, PhD and histologic characteristics as well as response to therapy. Lisa Christopher-Stine, MD, MPH Results: Three patients with anti–signal recognition particle (SRP) and 2 patients with anti-HMG- CoA reductase autoantibodies were included. The mean follow-up time after HiCy therapy was 37 6 28 months. Two patients demonstrated substantial response, evidenced by improved Correspondence to muscle strength and decreased muscle enzymes after HiCy therapy; both of these patients were Dr. Christopher-Stine: anti-SRP positive. Four patients experienced febrile neutropenia after HiCy therapy, one of which [email protected] required a prolonged intensive care unit stay for infectious complications, from which they even- tually recovered. Conclusions: These data suggest that HiCy therapy without stem cell rescue may be considered as an alternative for the treatment of refractory IMNM. Neurol Neuroimmunol Neuroinflamm 2017;4: e381; doi: 10.1212/NXI.0000000000000381

GLOSSARY ASCT 5 autologous stem cell transplantation; CK 5 creatine kinase; HiCy 5 high-dose cyclophosphamide; IG 5 immuno- globulin; IMNM 5 immune-mediated necrotizing myopathy; SRP 5 signal recognition particle.

Some patients with myositis may be resistant to standard immunosuppressive regiments, includ- ing prednisone, methotrexate, azathioprine, IV immunoglobulin (IG), rituximab, and/or other medications. In such patients, especially those with a progressive clinical course, additional ther- apies may be required. High-dose chemotherapy followed by stem cell rescue through autolo- gous stem cell transplantation (ASCT) has been previously used with some success in patients with refractory myositis.1–3 However, given the risks associated with ASCT, high- dose cyclophosphamide (HiCy) without bone marrow transplantation can also be considered. First used for treatment of severe aplastic anemia,4,5 HiCy has subsequently been applied to several rheumatologic and neurologic diseases.6–10 As HiCy seems to be better tolerated than ASCT and yet induces a high rate of remission in refractory autoimmune diseases,11 we have used it to treat a number of patients with refractory myositis at the Johns Hopkins Myositis Center.12 In this case series, we report the clinical response of 5 patients who underwent HiCy therapy for refractory immune-mediated necrotizing myopathy (IMNM), a particularly severe form of myositis that predominantly affects skeletal muscle.

METHODS This is a retrospective case series review of 5 patients with IMNM who were treated with HiCy at the Johns Hopkins Myositis Center. The Johns Hopkins Institutional Review Board approved this study. All patients had been evaluated as part of routine Supplemental data clinical care between 2009 and 2016. The patients were administered cyclophosphamide at a dose of 50 mg/kg of ideal body weight IV at Neurology.org/nn From the Johns Hopkins University School of Medicine (C.A.M., A.H.L., R.A.B., A.L.M., L.C.-S.), Baltimore; and National Institute of Health (A.L.M.), Bethesda, MD. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Table 1 Summary of demographic, clinical features, and immunosuppressive medications in 5 patients treated with HiCy

Age at Phenotype Follow-up period Patient diagnosis/sex/race (myositis-specific antibody) Previous treatments after HiCy therapy, mo

1 53/M/AA NM (anti-SRP) MEP, PDN, MTX, IVIG, AZA, MMF, TAC, corticotropin injection, RTX 34

2 33/F/C NM (anti-SRP) PDN, AZA, IVIG, RTX 72

3 36/M/AA NM (anti-SRP/anti-Ro) PDN, AZA, RTX 7

4 38/M/C NM (anti-HMGCR) PDN, AZA, IVIG, MTX, RTX, PLEX 59

5 22/F/AA NM (anti-HMGCR) PDN, MTX, MMF, IVIG, RTX, PLEX 14

Abbreviations: AA 5 African American; AZA 5 azathioprine; C 5 Caucasian; HiCy 5 high-dose cyclophosphamide; IG 5 immunoglobulin; MEP 5 methyl- prednisolone; MMF 5 mycophenolate mofetil; MTX 5 methotrexate; NM 5 necrotizing myopathy; PDN 5 prednisone; PLEX 5 plasmapheresis/plasma exchange; RTX 5 rituximab; SRP 5 signal recognition particle; TAC 5 tacrolimus.

every day for 4 consecutive days. The method of administering 9,000 IU/L, and he had muscle strength grade 3/5 in HiCy and the posttreatment supportive care has been previously proximal upper and lower extremities. Electromyog- 13 described. Bilateral arm abductor and hip flexor muscle strength raphy demonstrated an irritable myopathy, and was graded using a 0–5 modified Medical Research Council scale muscle biopsy showed severe necrotizing myopathy. and subsequently transformed to the 10-point Kendall scale and summed together for a score range of 0–40 as previously Lower extremity MRI showed atrophy and muscle described14 and included in the Supplementary Materials. This edema. Antibodies against signal recognition particle transformation has not been validated. Definite responders were (SRP) were detected in this patient’s serum. He was defined as those who both (1) improved by at least 2 grades in treated with corticosteroids and methotrexate either the upper or lower extremities and (2) were able to taper with minimal improvement. Azathioprine, myco- their prednisone dose below 10 mg/d. Nonresponders were phenolate mofetil, and IVIG treatment yielded tem- defined as patients who received therapy without improved muscle strength or ability to taper prednisone. Follow-up for porary improvement in muscle strength, but patients after HiCy administration was at the discretion of the worsening dysphagia and muscle pain. He was sub- primary treating rheumatologist and neurologist. Most patients sequently treated with ACTH gel (corticotropin were followed at 2–3-month intervals. injection), rituximab, and tacrolimus with no sub- stantial improvement in his symptoms. Given the RESULTS Table 1 summarizes demographic fea- refractory nature of his disease, the patient was treated tures, clinical characteristics, and immunosuppressive with HiCy. He tolerated the treatment well, but 9 therapies in 5 patients. Tables 2 and 3 summarize days after discharge, he was admitted to the hospital muscle strength, prednisone dose, and creatine kinase with neutropenic fever and treated for pneumonia. In (CK) level before and after HiCy treatment. Figure the ensuing 2 months, he experienced marked provides the overall change in strength over the first improvement in muscle strength, and his CK level year for each patient after receiving HiCy therapy. decreased from 2,000 to less than 500 IU/L. The Patient 1. A 53-year-old African American man pre- patient was completely weaned off immunosuppres- sented with a 4-month history of progressive upper sive therapy, and 3 month later, his muscle strength and lower extremity weaknesses, dyspnea on exertion, was graded 5/5 in both upper and lower extremities and dysphagia. His initial CK level was greater than and his CK level normalized. Patient 2. A 33-year-old white woman who developed

Table 2 Summary of creatine kinase (CK) value, glucocorticoid dose, and manual 5 months of progressive lower extremity weakness muscle strength testing before high-dose cyclophosphamide therapy and dyspnea was evaluated in clinic in March 2007. Her evaluation was notable for proximal muscle Max prednisone Deltoid Hip flexor Patient Peak CK, IU/L dose, mg/d strength, min strength, min weakness, CK of 8,000 IU/L, an irritable myopathy on electromyography, muscle biopsy showing a nec- Before therapy rotizing myopathy, and positive anti-SRP antibodies. Definite responders Her proximal upper extremities were graded as 4/5 1 9,000 60 3/5 3/5 and lower extremities 42/5. She was initially treated 2 8,495 60 42/5 2/5 with azathioprine and prednisone 60 mg daily with-

Nonresponders out effect. She subsequently received methotrexate

3 16,000 100 31/5 2/5 and IVIG (2 g/kg), but despite these therapies, she had persistent weakness, and her CK level remained 4 18,000 100 41/5 2/5 2,000–3,000 IU/L. She received rituximab with – – 5 22,050 80 1 2/5 1 2/5 suppression of her CD19/CD20 cell populations, CK normal range (24–170 IU/L). after which her CK actually increased to 6,000 IU/L.

2 Neurology: Neuroimmunology & Neuroinflammation Table 3 Summary of creatine kinase (CK) value, glucocorticoid dose, and manual muscle strength testing after high-dose cyclophosphamide therapy

Nadir Most recent Current prednisone Deltoid strength Hip flexor strength Patient CK, IU/L CK, IU/L dose, mg/d (at last visit) (at last visit)

After therapy

Definite responders

1 123 494 0 41/5 5/5

2 100 100 0 41/5 41/5

Nonresponders

3 1,058 3,594 10 42/5 2/5

4 245 714 10 5/5 3/5

5 663 3,923 40 every other day 1–2/5 1–2/5

CK normal range (24–170 IU/L).

She therefore received HiCy in November 2008. She a 40-lb unintentional weight loss. His evaluation re- tolerated the therapy well, but 9 days after the infu- vealed a CK level of 16,000 IU/L, and electromyog- sion, she was admitted to the hospital for neutropenic raphy showed an irritable myopathy. A muscle biopsy fever and received broad-spectrum antibiotics. No of his right thigh by report showed minimal inflam- infection was identified, and she subsequently mation. He was subsequently found to have both recovered. In January 2009, her strength was starting anti-SRP and anti-Ro autoantibodies. His proximal to progressively improve. Over the subsequent years, lower extremities were graded as 2/5 in strength she was able to taper her immunosuppressive regimen bilaterally. He was treated with high-dose prednisone, while retaining her strength and function. During her 100 mg/d, and azathioprine with initial improve- most recent follow-up appointment, her CK level ment, but this was transient, despite this therapy, he trended down to 100 IU/L, her strength had worsened clinically with a persistently elevated CK improved to 41/5 in proximal muscle groups, and level to ;4,000 IU/L and proximal muscle weakness she remained in clinical remission off of all immu- graded 2/5. He was subsequently trialed on metho- nosuppressive therapies. trexate and 4 doses of rituximab therapy with docu- mented suppressed CD19/CD20 cell counts. Patient 3. A 36-year-old African American man pre- However, he did not have any significant improve- sented to our clinic in February 2011 with a 7-month ment. He therefore was referred for and received history of proximal muscle weakness accompanied by HiCy in May 2016. His hospital course was com- plicated by neutropenic fever and fungal pneumonia as well as Clostridium difficile colitis. He was hospi- Figure Strength over time for each patient over the first year after receiving high-dose cyclophosphamide therapy talized for a total of 25 days. Within months, his CK level had decreased to 1,000 IU/L; however, at his most recent follow-up, 7 months after HiCy, his CK level again started to rise to 3,500 IU/L with a decrease in strength to pre-HiCy levels.

Patient 4. A 38-year-old white man presented with an 11-month history of progressive muscle weakness and myalgia. His initial evaluation was notable for a CK level of 18,000 IU/L, electromyography revealing an irritable myopathy and MRI of bilateral thighs demonstrating muscle edema of all compartments. A muscle biopsy sample showed severe necrotizing myopathy. Serum test revealed antibodies to 3- hydroxy-3-methylglutaryl-CoA reductase (HMGCR antibodies), with no history of statin use. The patient was started on prednisone 60 mg daily with worsen- ing muscle weakness in upper extremities grade 41/5 Y-axis is 0–40, where the deltoids and hip flexors were scored on a 0–5 scale, and sub- sequently transformed to a scale of 0–10 for each of the 4 muscles, resulting in a maximum and hip flexors grade 2/5. IVIG (2 g/kg) and azathi- score of 40. HiCy 5 high-dose cyclophosphamide. oprine treatments were administered, but the muscle

Neurology: Neuroimmunology & Neuroinflammation 3 weakness worsened. He was prescribed 2 doses of At her last visit in May 2016, approximately 1 year rituximab and methotrexate, resulting in slight after HiCy, she was noted to have persistent muscle improvement in muscle strength, but CK levels weakness (hip flexors 2/5) and still required 40 mg/d started to rise again within 6 months. Rituximab was prednisone every other day. started again, but it was accompanied by an ana- phylactic reaction. He then received 5 plasmapheresis TOXICITY All 5 patients tolerated the HiCy regimen treatments without any substantial improvement in well. All patients experienced mild side effects includ- strength. The patient was admitted and received ing nausea and alopecia. Neutropenic fever was – HiCy for 4 consecutive days in September 2011. He observed in 4 patients within 2 3 weeks after receiv- had 1 episode of neutropenic fever 15 days after ing HiCy, but all recovered. One patient (patient 3) HiCy, but he recovered successfully. Within 1 experienced a prolonged hospital stay of 25 days after month, strength remained at 4/5 in the proximal HiCy for treatment of C difficile colitis and fungal upper extremity and 3/5 in the hip flexors (weaker pneumonia. At the time of this report, no patient than the grades before HiCy therapy). The patient experienced hemorrhagic cystitis or malignancy. was started on mycophenolate mofetil without any There were no fatalities. improvement after 3 months. He underwent desen- DISCUSSION In this study, we have described our sitization for rituximab and received a dose of ritux- experience treating patients with severe refractory imab in intensive care unit. Five months after IMNM with HiCy without stem cell rescue. The 2 rituximab therapy, muscle strength improved to 5/5 definite responders (patients 1 and 2) had substantial in the upper extremities and 4/5 in the hip flexors, improvement in muscle strength and decreased and the CD19/CD20 counts remained suppressed. requirement for corticosteroids after treatment. Non- On his most recent visit, the patient required 10 mg responders included patients 3, 4, and 5 who had no prednisone daily with intermittent rituximab. improvement in strength or ability to taper immuno- Patient 5. A 22-year-old African American woman suppression. Patient 3 had initial improvement with first developed proximal muscle weakness 6 weeks subsequent return to pre-HiCy status. Patient 4 ini- after the delivery of her second child in January 2007. tially demonstrated partial response but then wors- She was evaluated at our clinic in March 2008 and ened, prompting treatment with rituximab, after was found to have a CK level of 15,000 IU/L and an which he demonstrated improvement. In all cases, the EMG showing an irritable myopathy. She underwent adverse effects were manageable and comparable with a muscle biopsy that revealed a necrotizing myopathy. those observed in previous HiCy studies.6,15 Although She was treated with 80 mg/d prednisone and 4 of the 5 patients required hospitalization for neu- methotrexate without substantial benefit. She was tropenic fever and/or infection, the high risk of this thus transitioned to IVIG, but her CK level actually side effect was anticipated and considered warranted, increased to over 17,000 IU/L. She then received given the limited additional treatment options and rituximab infusions with minimal improvement and the potential benefit of HiCy. This series suggests a nadir CK level of 4,800 IU/L. Mycophenolate was that HiCy may be effective in select patients with added with no appreciable difference. She also treatment-refractory IMNM. received 1 round of plasmapheresis, which resulted in IMNM is sometimes associated with unique auto- substantial improvement in strength but only lasted antibodies, including anti-SRP (patients 1–3) and a few days and was complicated by bacteremia and anti-HMGCR autoantibodies (patients 4–5), and pneumonia. Throughout this time, she was unable to a distinct clinical phenotype.16,17 Anti-HMGCR decrease her prednisone less than 20–30 mg/d. myopathy is characterized by prominent myofiber Serum test revealed antibodies to 3-hydroxy-3- necrosis on biopsy and very high CK levels.16,18 Sim- methylglutaryl-CoA reductase (HMGCR anti- ilarly, patients with anti-SRP antibodies frequently bodies), with no history of statin use. Given her present with markedly elevated CK levels and a nec- refractory nature and poor functional status requiring rotizing myopathy without primary inflamma- a wheelchair and family assistance, she received HiCy tion.17,19–21 Although it is interesting that both in April 2015. She tolerated the infusions well and patients who responded to HiCy were anti-SRP pos- did not have any complications after hospitalization. itive, the numbers included in this study are too small In May, she started to observe increasing proximal to conclude whether autoantibody status can predict muscle strength, and this continued throughout 2015 response to this therapy. (knee flexors/extensors by dynamometry increased Regarding the 3 patients with anti-SRP antibod- from 11–13 to 25–26 lbs bilaterally). Her CK level ies, 2 experienced dramatic improvement, whereas decreased to ;1,800 IU/L. However, in February the third had initial improvement with subsequent 2016, she was noted to start to decline functionally. worsening at the time of this report. Notably, the

4 Neurology: Neuroimmunology & Neuroinflammation patient who worsened also had anti-Ro autoantibod- R.A. Brodsky: drafting/revising manuscript. A.L. Mammen: drafting/ ies. Other groups have noted that in rheumatologic revising manuscript. L. Christopher-Stine: design/conceptualization and drafting/revising manuscript. conditions such as systemic lupus erythematosus and scleroderma, the presence of anti-Ro autoanti- STUDY FUNDING bodies portends a poorer prognosis22,23 and in myo- The Huayi and Siuling Zhang Discovery Fund. sitis has been associated with earlier and more severe disease.24–26 DISCLOSURE A review of the literature suggests that despite an C.A. Mecoli and A.H. Lahouti report no disclosures. R.A. Brodsky served on the scientific advisory board for Alexion, Achillion, and Apel- initial response rate, most patients with severe auto- lis; received research support from NIH, NHLBI, and AA/MDS Foun- immune disease ultimately relapse following HiCy dation; and is on the editorial board for Blood. A.L. Mammen served therapy with only 20% remaining disease free 5 years on the advisory board for aTYR Pharmaceuticals and Biogen; served on after treatment.12 However, it should be noted that the scientific advisory board for Experimental Neurology and Arthritis and Rheumatism; holds patent for anti-AMGCR antibodies; and patients with myositis were not included in those received research support from NIAMS/NIH and NIDDK/NIH. studies. We acknowledge that this therapy carries L. Christopher-Stine served on the scientific advisory board for Novar- a higher risk of infection and morbidity, especially tis, MedImmune, Walgreens/Option Care, Mallinckrodt, Idera Phar- maceuticals, Octapharma, and Corbus; received honoraria from when taking into account a potential high relapse rate Novartis, MedImmune, Walgreens/Option Care, Mallinckrodt, Idera as seen in other autoimmune conditions treated with Pharmaceuticals, and Octapharma; holds a patent for and receives roy- this regimen. Furthermore, long-term risk regarding alties from detection of HMG-CoA reductase (HMGCR) antibodies in malignancy must be taken into account, given its patient sera; consulted for NuFactor, Option Care, IgG America, Med- ProRx, CSL Behring, NHLBI; and participated in legal proceedings 27,28 association with some necrotizing myopathies. from McAloon and Friedman, P.C., Shaub, Ahmuty, Citrin, and However, for patients who have failed conventional Spratt, LLP, The Chartwell Law Offices, PK Law, and Bodie, Dolina, therapies for necrotizing myopathy (i.e., rituximab Hobbs, Friddell & Grenzer, P.C. Go to Neurology.org/nn for full and IVIG), we believe that this therapy offers another disclosure forms. option for the clinician to consider who otherwise has Received February 10, 2017. Accepted in final form May 30, 2017. very few options. Although follow-up of patients with myositis in REFERENCES the current study was limited to an average of 3 years 1. Henes JC, Heinzelmann F, Wacker A, et al. Antisignal (median 37 6 28 months, range 7–72 months), the 2 recognition particle-positive polymyositis successfully trea- responders have now been followed without relapse ted with myeloablative autologous stem cell transplanta- – for 34 and 72 months, respectively. Additional study tion. Ann Rheum Dis 2009;68:447 448. 2. Holzer U, van Royen-Kerkhof A, van der Torre P, et al. and follow-up time will be needed to further assess Successful autologous stem cell transplantation in two pa- the true duration of remission. tients with juvenile dermatomyositis. Scand J Rheumatol A strength of this study is that all 5 patients under- 2010;39:88–92. went the same HiCy protocol at a single center. In 3. BaronF,RibbensC,KayeO,FilletG,MalaiseM,BeguinY. addition, all have had longitudinal follow-up in the Effective treatment of Jo-1-associated polymyositis with same center. Limitations of the study include its ret- T-cell-depleted autologous peripheral blood stem cell transplantation. Br J Haematol 2000;110:339–342. rospective nature and the small number of patients. 4. Brodsky RA, Sensenbrenner LL, Jones RJ. Complete Furthermore, as this study was not protocol driven, remission in severe aplastic anemia after high-dose cyclo- follow-up was not uniform and at the discretion of phosphamide without bone marrow transplantation. the treating physician. Additionally, many patients Blood 1996;87:491–494. were treated with concurrent medications; however, 5. Brodsky RA, Chen AR, Dorr D, et al. High-dose cyclo- they were on stable doses for months preceding the phosphamide for severe aplastic anemia: long-term follow- up. Blood 2010;115:2136–2141. administration of HiCy. As such, the likelihood that 6. Moyo VM, Smith D, Brodsky I, Crilley P, Jones RJ, the patients had a delayed response to their back- Brodsky RA. High-dose cyclophosphamide for refrac- ground immunosuppressive medications is unlikely. tory autoimmune hemolytic anemia. Blood 2002;100: Although numerous therapies have been shown to 704–706. be effective for the treatment of inflammatory myopa- 7. Petri M, Brodsky RA, Jones RJ, Gladstone D, Fillius M, thies, treatment of refractory cases, particularly Magder LS. High-dose cyclophosphamide versus monthly IMNM, remains challenging.29 The data presented intravenous cyclophosphamide for systemic lupus erythe- matosus: a prospective randomized trial. Arthritis Rheum here reveal that HiCy may represent a potential ther- 2010;62:1487–1493. apeutic alternative in IMNM patients with refractory 8. Drachman DB, Adams RN, Hu R, Jones RJ, Brodsky RA. disease. Rebooting the immune system with high-dose cyclophos- phamide for treatment of refractory myasthenia gravis. AUTHOR CONTRIBUTIONS Ann NY Acad Sci 2008;1132:305–314. C.A. Mecoli: design/conceptualization and drafting/revising manuscript. 9. Henderson CF, Brodsky RA, Jones RJ, Levine SM. High- A.H. Lahouti: design/conceptualization and drafting/revising manuscript. dose cyclophosphamide without stem cell rescue for the

Neurology: Neuroimmunology & Neuroinflammation 5 treatment of refractory Behcet’s disease. Semin Arthritis A reductase in patients with statin-associated autoimmune Rheum 2011;41:301–304. myopathy. Arthritis Rheum 2011;63:713–721. 10. Krishnan C, Kaplin AI, Brodsky RA, et al. Reduction of 19. Miller T, Al-Lozi MT, Lopate G, Pestronk A. Myopathy disease activity and disability with high-dose cyclophos- with antibodies to the signal recognition particle: clinical phamide in patients with aggressive multiple sclerosis. and pathological features. J Neurol Neurosurg Psychiatry Arch Neurol 2008;65:1044–1051. 2002;73:420–428. 11. DeZern AE, Petri M, Drachman DB, et al. High-dose 20. Tiniakou E, Pinal-Fernandez I, Lloyd TE, Albayda J. cyclophosphamide without stem cell rescue in 207 patients More severe disease and slower recovery in younger pa- with aplastic anemia and other autoimmune diseases. tients with anti-3-hydroxy-3-methylglutaryl-coenzyme A Medicine (Baltimore) 2011;90:89–98. reductase-associated autoimmune myopathy. Rheuatology 12. Lahouti AH, Brodsky RA, Christopher-Stine L. Idiopathic (Oxford) Epub 2017 Jan 17. inflammatory myopathy treated with high-dose immunoa- 21. Pinal-Fernandez I, Parks C, Werner JL, et al. Longitudinal blative cyclophosphamide–a long-term follow-up study. course of disease in a large cohort of myositis patients with JAMA Neurol 2015;72:1205–1206. autoantibodies recognizing the signal recognition particle. 13. Brodsky RA, Sensenbrenner LL, Smith BD, et al. Durable Arthritis Care Res (Hoboken) 2017;69:263–270. treatment-free remission after high-dose cyclophospha- 22. Korbet SM, Schwartz MM, Evans J, Lewis EJ; Collabora- mide therapy for previously untreated severe aplastic ane- tive Study Group. Severe lupus nephritis: racial differences mia. Ann Intern Med 2001;135:477–483. in presentation and outcome. J Am Soc Nephrol 2007;18: 14. Rider LG, Werth VP, Huber AM, et al. Measures of adult 244–254. and juvenile dermatomyositis, polymyositis, and inclusion 23. Wodkowski M, Hudson M, Proudman S, et al. Mono- body myositis: physician and patient/parent global activity, specific anti-Ro52/TRIM21 antibodies in a tri-nation manual muscle testing (MMT), Health Assessment Ques- cohort of 1574 systemic sclerosis subjects: evidence of an tionnaire (HAQ)/Childhood Health Assessment Ques- association with interstitial lung disease and worse survival. tionnaire (C-HAQ), Childhood Myositis Assessment Clin Exp Rheumatol 2015;33:S131–S135. Scale (CMAS), Myositis Disease Activity Assessment Tool 24. Pinal-Fernandez I, Casal-Dominguez M, Huapaya JA, (MDAAT), Disease Activity Score (DAS), Short Form 36 et al. A longitudinal cohort study of the anti-synthetase (SF-36), Child Health Questionnaire (CHQ), Physician syndrome: increased severity of interstitial lung disease Global Damage, Myositis Damage Index (MDI), Quanti- in black patients and patients with anti-PL7 and anti- tative Muscle Testing (QMT), Myositis Functional Index- PL12 autoantibodies. Rheumatology (Oxford) 2017;56: 2 (FI-2), Myositis Activities Profile (MAP), Inclusion 999–1007. Body Myositis Functional Rating Scale (IBMFRS), Cuta- 25. La Corte R, Lo Mo Naco A, Locaputo A, Dolzani F, neous Dermatomyositis Disease Area and Severity Index Trotta F. In patients with antisynthetase syndrome the (CDASI), Cutaneous Assessment Tool (CAT), Dermato- occurrence of anti-Ro/SSA antibodies causes a more myositis Skin Severity Index (DSSI), Skindex, and Der- severe interstitial lung disease. Autoimmunity 2006; matology Life Quality Index (DLQI). Arthritis Care Res 39:249–253. (Hoboken) 2011;63(suppl 11):S118–S157. 26. Marie I, Hatron PY, Dominique S, et al. Short-term and 15. Tehlirian CV, Hummers LK, White B, Brodsky RA, long-term outcome of anti-Jo1-positive patients with Wigley FM. High-dose cyclophosphamide without stem cell anti-Ro52 antibody. Semin Arthritis Rheum 2012;41: rescue in scleroderma. Ann Rheum Dis 2008;67:775–781. 890–899. 16. Christopher-Stine L, Casciola-Rosen LA, Hong G, Chung 27. Allenbach Y, Keraen J, Bouvier AM, et al. High risk of T, Corse AM, Mammen AL. A novel autoantibody recog- cancer in autoimmune necrotizing myopathies: useful- nizing 200-kd and 100-kd proteins is associated with an ness of myositis specific antibody. Brain 2016;139: immune-mediated necrotizing myopathy. Arthritis Rheum 2131–2135. 2010;62:2757–2766. 28. Kadoya M, Hida A, Hashimoto Maeda M, et al. Cancer 17. Valiyil R, Casciola-Rosen L, Hong G, Mammen A, association as a risk factor for anti-HMGCR antibody-positive Christopher-Stine L. Rituximab therapy for myopathy asso- myopathy. Neurol Neuroimmunol Neuroinflamm 2016;3: ciated with anti-signal recognition particle antibodies: a case e290. doi: 10.1212/NXI.0000000000000290. series. Arthritis Care Res (Hoboken) 2010;62:1328–1334. 29. Albayda J, Christopher-Stine L. Novel approaches in the 18. Mammen AL, Chung T, Christopher-Stine L, et al. Auto- treatment of myositis and myopathies. Ther Adv Muscu- antibodies against 3-hydroxy-3-methylglutaryl-coenzyme loskelet Dis 2012;4:369–377.

6 Neurology: Neuroimmunology & Neuroinflammation Varicella zoster virus–infected cerebrovascular cells produce a proinflammatory environment

Dallas Jones, PhD ABSTRACT C. Preston Neff, PhD Objective: To test whether varicella zoster virus (VZV) infection of human brain vascular cells and Brent E. Palmer, PhD of lung fibroblasts directly increases proinflammatory cytokine levels, consistent with VZV as Kurt Stenmark, MD a causative agent in intracerebral VZV vasculopathy and giant-cell arteritis (GCA). Maria A. Nagel, MD Methods: Conditioned supernatant from mock- and VZV-infected human brain vascular adventi- tial fibroblasts (HBVAFs), human perineurial cells (HPNCs), human brain vascular smooth muscle

Correspondence to cells (HBVSMCs), and human fetal lung fibroblasts (HFLs) were collected at 72 hours postinfec- Dr. Nagel: tion and analyzed for levels of 30 proinflammatory cytokines using the Meso Scale Discovery [email protected] Multiplex ELISA platform. Results: Compared with mock infection, VZV infection led to significantly increased levels of the following: interleukin-8 (IL-8) in all cell lines examined; IL-6 in HBVAFs, HPNCs, and HFLs, with no change in HBVSMCs; and vascular endothelial growth factor A in HBVAFs, HBVSMCs, and HFLs, with a significant decrease in HPNCs. Other cytokines, including IL-2, IL-4, IL-15, IL-16, TGF-b, Eotaxin-1, Eotaxin-3, IP-10, MCP-1, and granulocyte macrophage colony-stimulating factor, were also significantly altered upon VZV infection in a cell type–specific manner. Conclusions: VZV infection of vascular cells can directly produce a proinflammatory environment that may potentially lead to prolonged arterial wall inflammation and vasculitis. The VZV-mediated increase in IL-8 and IL-6 is consistent with that seen in the CSF of patients with intracerebral VZV vasculopathy, and the VZV-mediated increase in IL-6 is consistent with the cytokine’s elevated levels in temporal arteries and plasma of patients with GCA. Neurol Neuroimmunol Neuroinflamm 2017;4:e382; doi: 10.1212/NXI.0000000000000382

GLOSSARY a-SMA 5 alpha–smooth muscle actin; BBB 5 blood-brain barrier; FBS 5 fetal bovine serum; FACS 5 fluorescence-activated cell sorting; GCA 5 giant-cell arteritis; GM-CSF 5 granulocyte macrophage colony-stimulating factor; HBVAF 5 human brain vascular adventitial fibroblast; HBVSMC 5 human brain vascular smooth muscle cell; HFL 5 human fetal lung fibroblast; HPNC 5 human perineurial cell; IFNg 5 interferon gamma; IL 5 interleukin; TARC 5 thymus and activation-regulated chemo- kine; TNF-a 5 tumor necrosis factor alpha; TNF-b 5 tumor necrosis factor beta; TGF-b 5 transforming growth factor b; VZV 5 varicella zoster virus; VEGF-A 5 vascular endothelial growth factor A.

Varicella zoster virus (VZV) vasculopathy is due to productive virus infection of intracerebral arteries leading to stroke or aneurysm,1,2 as supported by the presence of viral antigen, DNA, and herpesvirus particles in affected cerebral arteries of a patient with multiple infarcts3,4 and by the presence of VZV antigen in a basilar artery aneurysm from a patient who died of cardiac arrest and subarachnoid hemorrhage 2 months after occipital distribution zoster.5 Recent studies have expanded the spectrum of VZV vasculopathy to extracranial arteries. Indeed, VZV antigen was detected in 73/107 (70%) of temporal arteries from patients with giant-cell arteritis (GCA)6; in many of these arteries, VZV DNA and herpesvirus particles were also found.7 Analysis of cerebral and temporal arteries from patients with VZV vasculopathy has revealed loss of medial smooth muscle cells, a hyperplastic intima composed of cells expressing a-smooth

From the Department of Neurology (D.J., M.A.N.), Department of Medicine (C.P.N., B.E.P.), and Department of Pediatrics (K.S.), University of Colorado School of Medicine, Aurora. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by NIH AG032958. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 muscle actin, and disruption of the internal adjacent cells. Depending on the amount of initial VZV-infected elastic lamina.8 A striking and consistent fea- cells added and time, eventually all cells will be productively in- fected and die. Given the amount of virus in the inoculum used ture of VZV vasculopathy was arterial inflam- herein, at 72 hours of postinfection (hpi; height of cytopathic 1 mation, consisting mostly of CD4 and effect with 50–80% of cells infected), culture supernatants were CD81 T cells, as well as CD681 macrophages; collected and cells were harvested using sodium citrate15 to neutrophils were abundant in the arterial optimize detection of cell surface proteins by flow cytometry. adventitia during early but not late infection.9 Flow cytometry. Mock- and VZV-infected cells at 72 hpi were It is important that arterial inflammation was washed with fluorescence-activated cell sorting (FACS) buffer (phosphate-bufferedsaline containing 1% FBS) and stained with intimately associated with an overlying thick- R-phycoerythrin–conjugated mouse anti-human anti-VZV-gE ened intima, supporting the notion that (Millipore, Billerica, MA) antibody using the SiteClick Antibody inflammatory cells secrete soluble factors labeling kit (ThermoFisher, Waltham, MA) for 30 minutes at 4°C, washed with FACS buffer, and fixed with 1% para- (e.g., cytokines and matrix metalloproteinases) formaldehyde. Isotype controls were used in all stainings. Cells that contribute to vascular injury, remodeling, were analyzed using a Canto-II or LSR-II flow cytometer (BD and dysfunction.9,10 Although the presence of Immunocytometry Systems, San Jose, CA); .15,000 events were VZV in conjunction with inflammation has collected for all samples. Data were analyzed using Diva software (BD Biosciences) and FlowJo Software (Tree Star, Ashland, OR). been seen in both cerebral and temporal ar- teries in intracerebral VZV vasculopathy and Multiplex electrochemiluminescence immunoassay. In the liquid phase from cell culture supernatants, levels of interleukin GCA, respectively, VZV as the direct cause of (IL)-1b, IL-1a, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL- arterial inflammation has not been demon- 12p70, IL-12/23p40, IL-13, IL-15, IL-16, IL-17a, interferon strated definitively. Thus, we tested whether gamma (IFNg), tumor necrosis factor–alpha (TNF-a), tumor necrosis factor–beta (TNF-b), Eotaxin-1, Eotaxin-3, IFN-g– VZV infection induces proinflammatory cyto- induced protein 10 (IP-10), monocyte chemoattractant protein kines that result in arterial inflammation seen 1 (MCP1), MCP4, macrophage-derived chemokine (MDC), in VZV vasculopathy and GCA using 3 pri- macrophage inflammatory protein 1–alpha (MIP-1a), MIP-1b, mary human brain vascular cell lines: (1) thymus and activation-regulated chemokine (TARC), gran- ulocyte macrophage colony-stimulating factor (GM-CSF), and human brain vascular adventitial fibroblasts vascular endothelial growth factor–A (VEGF-A) were measured (HBVAFs), which are key regulators of vascu- using the Meso Scale Discovery (MSD) human cytokine 30-Plex lar tone, function, and inflammation11; (2) kit (Rockville, MD) according to the manufacturer’s instructions. human perineurial cells (HPNCs), the barrier Levels of transforming growth factor-b1 (TGF-b1) were mea- sured using the human TGF-b1 kit (MSD). For each cytokine, cells surrounding adventitial nerve bundles concentrations were calculated with reference to a standard curve that VZV must penetrate to infect adjacent derived using various concentrations of the standards assayed in vascular cells; and (3) human brain vascular the same manner as the cell culture supernatants. The lower limit of detection (LLOD) was calculated as the concentration of signal smooth muscle cells (HBVSMCs), which are that was 2.5 SD above the zero calibrator. The upper limit of 12 considered immunoprivileged but may detection was calculated as the concentration of signal that was change the phenotype in response to VZV 2.5 SD below the upper plateau of the standard curve. All samples infection and migrate to form the thickened were analyzed in duplicate from triplicate experiments. intima.8 Human fetal lung fibroblasts (HFLs) Statistical analysis. Statistical analysis was performed using served as control cells in this study. GraphPad Prism (GraphPad, San Diego, CA). Statistical signifi- cance was determined using the Student t test. METHODS Virus and cells. Primary HBVAFs, HPNCs (Sciencell, Carlsbad, CA), and HFLs (ATCC, Manassas, VA) RESULTS Mock-infected HBVAFs secrete the highest were seeded at 2,000 cells/cm2 in a basal fibroblast medium with basal level of cytokines compared with other mock- 2% fetal bovine serum (FBS), 1% fibroblast growth serum, and infected cells. Because adventitial fibroblasts in the 3 1% 100 penicillin-streptomycin (Sciencell). HBVSMCs outermost layer of arteries are key regulators of vascu- (Sciencell) were seeded at 2,000 cells/cm2 in a basal smooth lar tone, function, and inflammation,11 basal levels of muscle cell medium with 2% FBS, 1% smooth muscle cell growth serum, and 1% 1003 penicillin-streptomycin (Sciencell). cytokines in mock-infected HBVAFs were compared After 24 hours, the medium was changed to basal fibroblast or with those in other mock-infected vascular cells basal smooth muscle cell medium with 0.1% FBS and 1% 1003 (HPNCs and HBVSMCs) and in mock-infected penicillin-streptomycin that was replenished every 48–72 hours HFLs commonly used to study VZV pathogenesis. – for 6 7 days to establish quiescence. At day 7, quiescent As compared to mock-infected HPNCs, HBVSMCs, HBVAFs, HPNCs, HBVSMCs, and HFLs were cocultivated and HFLs (figure 1; red, blue, and green bars, with VZV-infected (30–40 pfu/mL; Ellen strain)13,14 or unin- fected (mock-infected) HBVAFs, HPNCs, HBVSMCs, or HFLs, respectively), mock-infected HBVAFs showed sig- respectively. During the initial phase of cocultivation, a portion of nificantly higher levels of IL-2 (p , 0.005 for all); IL- cells in culture is infected; as infection progresses, virus spreads to 6(p , 0.0005, p , 0.005, and p , 0.0005,

2 Neurology: Neuroimmunology & Neuroinflammation Figure 1 HBVAFs secrete the highest level of cytokines among vascular and lung cells

Cell culture supernatants from mock-infected HBVAFs, HPNCs, HBVSMCs, and HFLs were harvested at 72 hpi and ana- lyzed for IL-1b, IL-1a, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-10, IL-12p70, IL-12/23p40, IL-13, IL-15, IL-16, IL-17a, IFNg, TNF-a, TNF-b, Eotaxin-1, Eotaxin-3, IP-10, MCP-1, MCP-4, MDC, MIP-1a, MIP-1b, TARC, GM-CSF, TGF-b, and VEGF-A levels. Only significant fold-change in HBVAFs cytokine levels compared with HPNCs (red), HBVSMCs (blue), or HFLs (green) were plotted. Bar graphs represent mean 6 SEM fold-change in cytokine levels from triplicate experiments. (a) p , 0.05, (b) p , 0.005, and (c) p , 0.0005. GM-CSF 5 granulocyte macrophage colony-stimulating factor; HBVAF 5 human brain vascular adventitial fibroblast; HBVSMC 5 human brain vascular smooth muscle cell; HFL 5 human fetal lung fibroblast; HPNC 5 human perineurial cell; IL 5 interleukin; TNF-a 5 tumor necrosis factor alpha; TNF-b 5 tumor necrosis factor beta; TARC 5 thymus and activation-regulated chemokine; TGF-b 5 transforming growth factor b; VEGF-A 5 vascular endothelial growth factor A. respectively); IL-8 (p , 0.0005 for all); IL-10 (p , infected HBVAFs had significantly increased levels of 0.005 for all); IL-12p70 (p , 0.0005, p , 0.005, and IL-8 (p 5 0.02), IL-6 (p 5 0.03), VEGF-A p , 0.0005); IL-13 (p , 0.0005, p , 0.005, and p , (p 5 0.001), TGF-b (p 5 0.04), and IL-16 (p 5 0.0005); IL-15 (p , 0.0005 for all); TNF-a (p , 0.01), along with significantly decreased levels of 0.005, p , 0.05, and p , 0.0005, respectively); Eotaxin-1 (p 5 0.002), MCP-1 (p 5 0.003), IP-10 IFN-g (p , 0.0005, p , 0.005, and p , 0.0005, (p 5 0.02), and GM-CSF (p 5 0.03) (table and respectively); GM-CSF (p , 0.0005, p , 0.0005, figure 2). Levels of the remaining 21 analytes were and p , 0.005, respectively); and VEGF-A (p , either undetectable or not significantly changed. 0.005, p , 0.0005, and p , 0.0005, respectively), whereas TGF-b levels were significantly decreased Specific cytokines significantly altered during VZV (p , 0.05, p , 0.0005, and p , 0.005, respectively). infection of HPNCs. VZV-infected HPNCs at 72 hpi . 1 Unique differences in HBVAF cytokine levels com- were 85% VZV gE . Compared with the super- pared with those of HPNCs included significant in- natant from mock-infected HPNCs, the supernatant creases in IL-1b (p , 0.005), IL-4 (p , 0.005), and from VZV-infected HPNCs had significantly 5 5 IP-10 (p , 0.005), but a significant decrease in increased levels of IL-8 (p 0.05), IL-6 (p 0.006), 5 5 MCP-1 (p , 0.005). Unique differences in HBVAF IL-2 (p 0.03), and GM-CSF (p 0.02), along 5 cytokine levels compared with HBVSMCs showed with significantly decreased levels of VEGF-A (p significant increases in Eotaxin-1 (p , 0.005), IP-10 0.003) (table and figure 3). Levels of the remaining (p , 0.05), and TARC (p , 0.05), but a significant 25 analytes were either undetectable or not signifi- decrease in Eotaxin-3 (p , 0.05), whereas the com- cantly changed. parison with HFL cytokine levels revealed showed Specific cytokines significantly altered during VZV significant increases in IL-1b (p , 0.005) and IL-4 infection of HBVSMCs. VZV-infected HBVSMCs at (p , 0.005). 72 hpi were .50% VZV gE1. Compared with the Specific cytokines significantly altered during VZV supernatant from mock-infected HBVSMCs, the infection of HBVAFs. To monitor VZV infection, supernatant from VZV-infected HBVSMCs had mock- and VZV-infected HBVAFs were harvested significantly increased levels of IL-8 (p 5 0.002) and at 72 hpi and analyzed for VZV gE expression using VEGF-A (p 5 0.0004), along with significantly flow cytometry. VZV-infected HBVAFs were .80% decreased levels of IL-15 (p 5 0.006), Eotaxin-3 (p 5 VZV gE1. Compared with the supernatant from 0.004), IP-10 (p 5 0.02), and MCP-1 (p 5 0.02) mock-infected HBVAFs, the supernatant from VZV- (table and figure 4). Levels of the remaining 24

Neurology: Neuroimmunology & Neuroinflammation 3 Table Meso Scale Discovery analyte levels from mock- and VZV-infected HBVAFs, HPNCs, HBVSMCs, and HFLs

HBVAFs, Mock/VZV HPNCs, Mock/VZV HBVSMCs, Mock/VZV HFLs, Mock/VZV LLOD–ULOD

IL-1b 0.372 6 0.100 0.131 6 0.078 0.268 6 0.110 0.083 6 0.067 0.140–575.0

0.221 6 0.123 0.150 6 0.099 0.211 6 0.094 0.175 6 0.092

IL-1a Undetectable Undetectable Undetectable Undetectable 0.0989–405.0

Undetectable Undetectable 0.101 6 0.084 Undetectable

IL-2 1.09 6 0.183 0.458 6 0.044a 0.492 6 0.123 0.221 6 0.068 0.381–1,560.0

1.18 6 0.175 0.624 6 0.083a 0.703 6 0.202 0.477 6 0.237

IL-4 0.250 6 0.059 0.055 6 0.016 0.159 6 0.04 0.038 6 0.015a 0.0583–239.0

0.323 6 0.045 0.084 6 0.017 0.198 6 0.048 0.063 6 0.016a

IL-5 Undetectable Undetectable Undetectable Undetectable 0.223–912.0

Undetectable Undetectable Undetectable Undetectable

IL-6 437.7 6 32.85a 59.44 6 2.4a 315.2 6 23.0 44.8 6 4.9a 0.182–745.0

630.5 6 45.72a 74.7 6 5.0a 275.3 6 22.5 74.6 6 10.7a

IL-7 Undetectable Undetectable Undetectable Undetectable 0.205–838.0

Undetectable Undetectable Undetectable Undetectable

IL-8 546.9 6 38.5a 305.4 6 16.0a 158.8 6 11.9a 140.8 6 8.0a 0.163–666.0

695.1 6 56.6a 345.3 6 18.9a 234.8 6 9.2a 223.1 6 26.3a

IL-10 0.366 6 0.061 0.136 6 0.059 0.135 6 0.038 0.078 6 0.053 0.0891–365.0

0.369 6 0.109 0.168 6 0.028 0.207 6 0.075 0.109 6 0.041

IL-12p70 1.313 6 0.096 0.303 6 0.083 0.736 6 0.20 0.152 6 0.095 0.119–489.0

1.414 6 0.235 0.411 6 0.038 0.741 6 0.234 0.188 6 0.093

IL-12/23p40 Undetectable Undetectable Undetectable Undetectable 0.786–3,220.0

Undetectable Undetectable Undetectable Undetectable

IL-13 3.488 6 0.184 1.287 6 0.224 1.798 6 0.476 0.511 6 0.331 0.123–505.0

4.089 6 0.454 2.143 6 0.212 2.271 6 0.693 1.074 6 0.380

IL-15 1.483 6 0.071 0.154 6 0.273 0.783 6 0.013b 0.802 6 0.035a 0.190–779.0

1.482 6 0.099 0.322 6 0.395 0.504 6 0.044b 1.135 6 0.157a

IL-16 Undetectablea Undetectable Undetectable Undetectable 0.618–2,530.0

1.177 6 0.824a Undetectable Undetectable Undetectable

IL-17a Undetectable Undetectable Undetectable Undetectable 1.47–6,010.0

Undetectable Undetectable Undetectable Undetectable

TNF-a 0.537 6 0.06 0.212 6 0.079 0.337 6 0.09 0.10 6 0.045 0.0901–369.0

0.601 6 0.11 0.309 6 0.058 0.442 6 0.12 0.233 6 0.099

TNF-b Undetectable Undetectable Undetectable Undetectable 0.154–632.0

Undetectable Undetectable Undetectable Undetectable

TGF-b 132.3 6 28.9a 352.5 6 92.9 498.8 6 42.4 222.7 6 22.3a 17.0–100,000.0

256.7 6 66.5a 337.8 6 36.1 532.4 6 92.8 786.2 6 193.1a

IFN-g 3.049 6 0.341 0.677 6 0.40 1.64 6 0.41 Undetectable 0.396–1,620.0

3.221 6 0.887 1.005 6 0.248 1.871 6 0.90 Undetectable

Eotaxin-1 13.56 6 2.81b 9.98 6 4.13 2.72 6 2.70 10.51 6 7.91 0.366–1,500.0

1.174 6 1.758b 8.82 6 2.82 Undetectable 7.26 6 2.0

Eotaxin-3 3.257 6 2.144 3.917 6 1.953 8.063 6 2.154b 2.413 6 2.224 1.21–4,970.0

1.877 6 1.412 6.687 6 3.303 3.62 6 1.869b 3.369 6 1.806

IP-10 2.418 6 0.546b 0.771 6 0.177 1.007 6 0.434b 1.312 6 0.693 0.586–2,400.0

0.425 6 0.103b 1.031 6 0.181 0.308 6 0.073b 0.794 6 0.249

MCP-1 142.3 6 11.05b 445.2 6 96.32 130.8 6 25.9b 180.4 6 69.17 0.109–448.0

Continued

4 Neurology: Neuroimmunology & Neuroinflammation Table Continued

HBVAFs, Mock/VZV HPNCs, Mock/VZV HBVSMCs, Mock/VZV HFLs, Mock/VZV LLOD–ULOD

58.11 6 18.40b 356.6 6 90.12 71.19 6 13.59b 210 6 71.95

MCP-4 16.79 6 5.81 29.26 6 10.0 16.49 6 5.81 20.65 6 7.95 0.147–602.0

19.64 6 3.45 29.30 6 3.58 12.49 6 7.25 24.17 6 6.60

MDC Undetectable 4.085 6 2.20 Undetectable Undetectable 2.59–10,600.0

Undetectable 2.985 6 0.66 Undetectable Undetectable

MIP-1a Undetectable 8.345 6 3.69 Undetectable Undetectable 0.230–942.0

Undetectable 4.834 6 3.16 Undetectable Undetectable

MIP-1b Undetectable Undetectable Undetectable Undetectable 0.247–1,010.0

Undetectable Undetectable Undetectable Undetectable

TARC 3.068 6 0.973 6.005 6 4.93 1.649 6 0.311 6.369 6 5.774 0.366–1,500.0

3.689 6 3.66 3.831 6 1.783 3.494 6 2.456 2.68 6 1.03

GM-CSF 14.54 6 0.46b 0.06 6 0.10a 4.001 6 0.378 11.17 6 4.17 0.243–996.0

12.17 6 1.04b 2.52 6 1.16a 3.953 6 0.190 10.19 6 2.43

VEGF-A 310.9 6 16.3a 152.7 6 52.5b 17.7 6 4.59a 94.5 6 26.3a 0.249–1,020.0

523.4 6 23.6a 9.58 6 2.9b 127.4 6 3.6a 366.9 6 108.7a

Abbreviations: GM-CSF 5 granulocyte macrophage colony-stimulating factor; HBVAF 5 human brain vascular adventitial fibroblast; HBVSMC 5 human brain vascular smooth muscle cell; HFL 5 human fetal lung fibroblast; HPNC 5 human perineurial cell; IL 5 interleukin; LLOD 5 lower limit of detection; ULOD 5 upper limit of detection; TARC 5 thymus and activation-regulated chemokine; TGF-b 5 transforming growth factor b; TNF-a 5 tumor necrosis factor alpha; TNF-b 5 tumor necrosis factor beta; VEGF-A 5 vascular endothelial growth factor A; VZV 5 varicella zoster virus. a Significantly increased cytokine levels in VZV-infected vs mock-infected cells. b Significantly decreased cytokine levels in VZV-infected vs mock-infected cells.

analytes were either undetectable or not significantly and IL-16, were altered in response to VZV infection changed. and warrant further discussion. For example, IL-8 (CXCL8), which promotes neutrophil migration to Specific cytokines significantly induced during VZV the site of infection and activation through degranu- infection of HFLs. As a positive control for cytokine lation,16 was the only cytokine/chemokine elevated in secretion during VZV infection, we analyzed cytokine all 4 cell lines studied. This finding is consistent with levels in mock- and VZV-infected HFLs. VZV- a significant increase of IL-8 levels in the CSF of infected HFLs at 72 hpi were .80% VZV gE1. patients with VZV vasculopathy compared with Compared with mock-infected HFLs, VZV-infected CSF from both patients with MS and healthy con- HFLs had significantly increased levels of IL-8 (p 5 trols.17 Moreover, robust IL-8 induction explains the 0.02), IL-6 (p 5 0.02), VEGF-A (p 5 0.01), TGF-b neutrophil infiltration seen in VZV-infected temporal (p 5 0.01), IL-15 (p 5 0.04), and IL-4 (p 5 0.05) arteries9 and the predominance of neutrophils in CSF (table and figure 5). Levels of the remaining 24 an- of patients with VZV vasculopathy.18 Finally, on dis- alytes were either undetectable or not significantly continuation of corticosteroids in patients with GCA, changed. serum IL-8 increases and neutrophils are activated,19 DISCUSSION Herein, we show that VZV infection strongly suggesting that the underlying cause of of primary HPNCs, brain vascular adventitial fibro- GCA, i.e., VZV, is still present to trigger arterial blasts and vascular smooth muscle cells, as well as inflammation. lung fibroblasts significantly upregulates multiple Levels of IL-6, an acute-phase proinflammatory proinflammatory cytokines, consistent with the cytokine that promotes monocyte differentiation into notion that VZV infection of vascular cells can initi- macrophages,20 were increased in VZV-infected in ate infiltration of inflammatory cells, vasculopathy, HPNCs, HBVAFs, and HFLs, as also seen in the and ultimately stroke or GCA. Thus, the association CSF of patients with VZV vasculopathy.17 Indeed, of VZV and inflammation is unlikely to reflect preex- macrophages represent a dominant immune infiltrate isting inflammation triggering VZV reactivation; in VZV-infected arteries from patients with intrace- rather, VZV directly contributes to vasculitis. rebral VZV vasculopathy, as well as GCA.7,9 In Several proinflammatory cytokines important in HBVSMCs, the lack of a VZV-induced increase in persistent inflammation and vascular remodeling, IL-6, as well as decreases in IL-15, Eotaxin-3, IP-10, i.e., IL-8, IL-6, GM-CSF, VEGF-A, TGF-b, IL-15, and MCP-1, adds another contributing factor to the

Neurology: Neuroimmunology & Neuroinflammation 5 Figure 2 Specific cytokines significantly altered during VZV infection in HBVAFs

Cells and cell culture supernatants were harvested at 72 hpi from mock- or VZV-infected HBVAFs. (A) Flow cytometry analyses of VZV gE expression in mock- (blue) and VZV-infected (red) HBVAFs at 72 hpi showed that .80% of cells were VZV gE1. Cells were gated using isotype controls. Compared with cell culture supernatants from mock-infected cells, supernatants from VZV-infected HBVAFs had significantly higher levels of IL-8 (B; p 5 0.02), IL-6 (C; p 5 0.03), VEGF-A (D; p 5 0.001), TGF-b (E; p 5 0.04), and IL-16 (F; p 5 0.01). Compared with cell culture supernatants from mock-infected cells, supernatants from VZV-infected HBVAFs had significantly lower levels of Eotaxin-1 (G; p 5 0.002), MCP-1 (H; p 5 0.003), IP-10 (I; p 5 0.02), and GM- CSF (J; p 5 0.03). All cytokine values are given in picograms per milliliter. Bar graphs represent mean 6 SD cytokine levels from 3 independent experiments. GM-CSF 5 granulocyte macrophage colony-stimulating factor; HBVAF 5 human brain vascular adventitial fibroblast; IL 5 interleukin; TGF-b 5 transforming growth factor b; VEGF-A 5 vascular endothelial growth factor A; VZV 5 varicella zoster virus.

immunoprivileged status of VSMCs.12 Previous stud- The lack of IL-6 induction in VZV-infected ies have shown that smooth muscle cells intrinsically HBVSMCs might also represent a protective host lack essential costimulators, particularly OX40L, that mechanism because IL-6 induces smooth muscle cell activate memory T cells.21 Our finding that the proin- apoptosis.22 flammatory response to virus infection is also reduced Two other cytokines involved in macrophage mat- in smooth muscle cells may contribute to the lack of uration and activation were differentially regulated in medial inflammation during early VZV infection.6 VZV-infected HBVAFs and HPNCs. GM-CSF

Figure 3 Specific cytokines significantly altered during VZV infection in HPNCs

Cells and cell culture supernatants were harvested at 72 hpi from mock- or VZV-infected HPNCs. (A) Flow cytometry analyses of VZV gE expression in mock- (blue) and VZV-infected (red) HPNCs at 72 hpi showed that .85% of cells were VZV gE1. Cells were gated using isotype controls. Compared with cell culture supernatants from mock-infected cells, supernatants from VZV-infected HPNCs had significantly higher levels of IL-8 (B; p 5 0.05), IL-6 (C; p 5 0.006), IL-2 (D; p 5 0.03), and GM-CSF (E; p 5 0.02). Compared with cell culture supernatants from mock-infected cells, supernatants from VZV-infected HPNCs had significantly lower levels of VEGF-A (F; p 5 0.003). All cytokine values are given in picograms per milliliter. Bar graphs represent mean 6 SD cytokine levels from 3 independent experiments. GM-CSF 5 granulocyte macrophage colony-stimulating factor; HPNC 5 human perineurial cell; IL 5 interleukin; VEGF-A 5 vascular endothelial growth factor A; VZV 5 varicella zoster virus.

6 Neurology: Neuroimmunology & Neuroinflammation protein for macrophages and neutrophils.23 The role Figure 4 Cytokines significantly altered during VZV infection in HBVSMCs of VEGF-A in stroke is contextual and can be both beneficial and problematic.23 The benefits of VEGF- A include neuronal protection, vasodilation, and angiogenesis; however, high levels of VEGF-A can increase blood-brain barrier (BBB) leakage to produce poststroke brain edema and severe intracranial hypertension.24,25 Patients with high basal levels of VEGF-A have a greater risk of stroke.26–28 In addi- tion, analyses in murine models of stroke have shown that prompt administration of VEGF-A increased stroke severity, whereas prompt VEGF-A blockade reduced infarct size and swelling.24,25 Thus, the VZV-mediated induction of VEGF-A in HBVAFs and HBVSMCs would worsen the severity of VZV vasculopathy because these infected cells in the artery would be a continual source of VEGF-A acting to disrupt the BBB, enhance intracranial hypertension, and promote macrophage/neutrophil infiltration. Note that simian varicella virus infection induced VEGF-A, along with IL-8 and IL-6, during acute 29 Cells and cell culture supernatants were harvested at 72 hpi from mock- or VZV-infected infection in rhesus macaques, further supporting HBVSMCs. (A) Flow cytometry analyses of VZV gE expression in mock- (blue) and VZV-in- the VZV-mediated induction of VEGF-A reported fected (red) HBVSMCs at 72 hpi showed that .50% of cells were VZV gE1. Cells were herein. Although diminished VEGF-A secretion by gated using isotype controls. Compared with cell culture supernatants from mock-infected VZV-infected HPNCs in the artery could initially cells, supernatants from VZV-infected HBVSMCs had significantly higher levels of IL-8 (B; p 5 0.002) and VEGF-A (C; p 5 0.0004). Compared with cell culture supernatants from mock- prevent these adverse effects of VEGF-A, virus is infected cells, supernatants from VZV-infected HBVSMCs had significantly lower levels of not cleared from this cellular layer during reactivation IL-15 (D; p 5 0.006), Eotaxin-3 (E; p 5 0.004), IP-10 (F; p 5 0.02), and MCP-1 (G; p 5 0.02). and can spread to the adventitia and smooth muscle 6 All cytokine values are given in picograms per milliliter. Bar graphs represent mean SD cell layer. Diminished VEGF-A secretion by VZV- cytokine levels from 3 independent experiments. HBVSMC 5 human brain vascular smooth muscle cell; IL 5 interleukin; VEGF-A 5 vascular endothelial growth factor A; VZV 5 varicella infected HPNCs might reflect the delay in neuro- zoster virus. nal/arterial protection required for viral spread and VZV vasculopathy onset. levels were increased in VZV-infected HPNCs, pro- Both VZV-infected HBVAFs and HFLs showed moting macrophage maturation in conjunction with significantly increased secretion of TGF-b, a pleo- IL-6, whereas infected HBVAFs had decreased tropic cytokine that can induce migration and pro- GM-CSF, as well as decreased MCP-1, which would liferation.30 This increase in TGF-b mediated by diminish macrophage maturation/activation during VZV could enhance the migration of adventitial fi- the later stages of infection compared with HPNCs. broblasts in the infected artery, promoting the vascu- Perineurial cells are the first cells infected by VZV lar remodeling that is a hallmark of stroke/vasculitis. during reactivation, and their enhanced levels of TGF-b is also an immunosuppressive cytokine that GM-CSF and IL-6 during infection could be the ini- would diminish viral clearance in the infected artery tial stimulus to activate and differentiate monocytes and possibly explain our recent finding of VZV- into macrophages and retain them in the infected mediated downregulation of major histocompatibility artery. As virus spreads to the adventitia without complex-I in uninfected bystander cells.31 proper clearance, VZV-infected HBVAFs secrete less Along with TGF-b, IL-16, a cytokine that acti- GM-CSF, potentially preventing macrophage activa- vates CD41 T cells and can be secreted by TGF-b– tion and explaining the presence of macrophages in stimulated fibroblasts,32 was also significantly elevated arteries up to 10 months after diagnosis of VZV in VZV-infected HBVAFs, consistent with the infil- vasculopathy. trate of CD41 T cells observed in both early and late The most significantly altered secretory protein VZV vasculopathy. Significant elevations in IL-2 observed in all 4 cell lines analyzed was VEGF-A, secretion seen in VZV-infected HPNCs would also with levels significantly increased in VZV-infected aid in T-cell activation. Of interest, VZV has a tro- HBVAFs, HBVSMCs, and HFLs, but significantly pism for CD41 T cells, and the induction of IL-2 and decreased in infected HPNCs. VEGF-A is of particu- IL-16 in the infected arteries could aid in the viral lar importance during VZV vasculopathy as it pro- tropism for these cells. Both VZV-infected HBVAFs motes angiogenesis, along with being a chemotactic and HBVSMCs also secreted significantly less IP-10,

Neurology: Neuroimmunology & Neuroinflammation 7 hallmarks of this disease, along with pulmonary hyper- Figure 5 Cytokines significantly induced during VZV infection in HFLs tension and stroke in general.11 Overall, our comprehensive analysis of the cyto- kines secreted by multiple cells in the artery and by lung fibroblasts after VZV infection shows that the virus significantly upregulates and downregulates sev- eral cytokines in various combinations distinct for each cell type. Most notable in VZV-infected vascular cells is the detection of elevated IL-8 and IL-6, which promote neutrophil and macrophage migra- tion/activation in the infected artery, as well as the detection of elevated VEGF-A, which promotes BBB permeability and immune infiltration. Of inter- est, IL-6 is the major cytokine involved in GCA path- ogenesis,38–40 whereas IL-8 and VEGF-A have been found to play a major role in disease progression as well.39 Together with a previous report that VZV- infected arteries downregulate PD-L1 to promote persistent inflammation,31 these data demonstrate that VZV infection is sufficient to promote a proin- flammatory environment within 72 hours, which may potentially lead to a persistent vasculitis, and Cells and culture supernatants were harvested at 72 hpi from mock- or VZV-infected HFLs. support VZV as the causative agent in intracerebral (A) Flow cytometry analyses of VZV gE expression in mock- (blue) and VZV-infected (red) HFLs at 72 hpi showed that .80% of cells were VZV gE1. Cells were gated using isotype VZV vasculopathy and GCA. controls. Compared with cell culture supernatants from mock-infected cells, supernatants from VZV-infected HFLs had significantly higher levels of IL-8 (B; p 5 0.02), IL-6 (C; p 5 AUTHOR CONTRIBUTIONS 0.02), VEGF-A (D; p 5 0.01), TGF-b (E; p 5 0.01), IL-15 (F; p 5 0.04), and IL-4 (G; p 5 0.05). All Dr. Jones: drafted and revised the manuscript for content, and collected, cytokine values are given in pg/ml. Bar graphs represent mean 6 SD cytokine levels from 3 analyzed, and interpreted the data. Dr. Neff: assisted with sample analysis independent experiments. HFL 5 human fetal lung fibroblast; IL 5 interleukin; TGF-b 5 and drafted and revised the manuscript. Dr. Palmer: assisted with sample transforming growth factor b; VEGF-A 5 vascular endothelial growth factor A; VZV 5 var- analysis and drafted and revised the manuscript. Dr. Stenmark: assisted icella zoster virus. with sample analysis and drafted and revised the manuscript. Dr. Nagel: drafted and revised the manuscript for content; designed and supervised the study; and collected, analyzed, and interpreted the data. which could inhibit viral clearance in the infected artery because IP-10 acts as a chemoattractant for ACKNOWLEDGMENT activated T cells and is required for host defense The authors thank Marina Hoffman for editorial review and Cathy Allen against virus-induced neurologic disorders.33 The for word processing and formatting. VZV-mediated targeting of this pathway to promote viral spread and persistent inflammation during VZV STUDY FUNDING This work was supported in part by the NIH grants NS094758 and vasculopathy is supported by analysis of IP-10 murine AG032958 to M.A.N. knockout models, which revealed a dramatic reduc- tion in viral clearance and enhanced inflammation DISCLOSURE during infection of the eye with herpes simplex-1, D. Jones and C. Preson Neff report no disclosures. B.E. Palmer received another member of the alpha-herpesviridae subfam- research support from NIH/NIDDK and NHLBI. K. Stenmark served ily, compared with wild-type mice.34 on the scientific advisory board for Pfizer, Entelligence/Actelion, and ContraFect; served on the editorial board for American Journal of Respi- IL-15 is a proinflammatory cytokine that activates ratory and Critical Care Medicine, American Journal of Physiology—Lung 1 natural killer cells and enhances memory CD8 T-cell Cellular and Molecular Physiology, Circulation Research, Pulmonary Circu- responses.35 Based on the functional properties of this lation, American Journal of Respiratory Cell and Molecular Biology; con- sulted for Novartis and ContraFect; and received research support from cytokine, the VZV-mediated induction of IL-15 in NIH, Axis, and Department of Defense. M.A. Nagel received research HFLs could enhance viral clearance in the lung during support from NIH. Go to Neurology.org/nn for full disclosure forms. primary infection. VZV-infected HBVSMCs had sig- nificantly reduced levels of secreted IL-15, which has Received January 9, 2017. Accepted in final form May 16, 2017. been shown to inhibit smooth muscle cell prolifera- tion.36 Furthermore, blockade of IL-15 during arterial REFERENCES 1. Nagel MA, Cohrs RJ, Mahalingam R, et al. The varicella injury increased intimal thickening in mice,37 afinding zoster virus vasculopathies: clinical, CSF, imaging, and particularly relevant in VZV vasculopathy because virologic features. Neurology 2008;70:853–860. a thickened intima and smooth muscle cell migration 2. Liberman AL, Nagel MA, Hurley MC, Caprio FZ, and differentiation into a myofibroblast phenotype are Bernstein RA, Gilden D. Rapid development of 9 cerebral

8 Neurology: Neuroimmunology & Neuroinflammation aneurysms in varicella-zoster virus vasculopathy. Neurol- for neutrophils in giant-cell arteritis disease progress. Circ ogy 2014;82:2139–2141. Res 2014;114:242–248. 3. Case records of the Massachusetts General Hospital. 20. Chomarat P, Banchereau J, Davoust J, Palucka AK. IL-6 Weekly clinicopathological exercises. Case 5-1995. A 73- switches the differentiation of monocytes from dendritic year-old man with focal brain lesions and peripheral-nerve cells to macrophages. Nat Immunol 2000;1:510–514. disease. N Engl J Med 1995;332:452–459. 21. Zhang P, Manes TD, Pober JS, Tellides G. Human vas- 4. Gilden DH, Kleinschmidt-DeMasters BK, Wellish M, cular smooth muscle cells lack essential costimulatory mol- Hedley-Whyte ET, Rentier B, Mahalingam R. Varicella ecules to activate allogeneic memory T cells. Arterioscler zoster virus, a cause of waxing and waning vasculitis: the Thromb Vasc Biol 2010;30:1795–1801. New England Journal of Medicine case 5-1995 revisited. 22. Yu H, Clarke MC, Figg N, Littlewood TD, Bennett MR. Neurology 1996;47:1441–1446. Smooth muscle cell apoptosis promotes vessel remodeling 5. Fukumoto S, Kinjo M, Hokamura K, Tanaka K. Sub- and repair via activation of cell migration, proliferation, arachnoid hemorrhage and granulomatous angiitis of the and collagen synthesis. Arterioscler Thromb Vasc Biol basilar artery: demonstration of the varicella-zoster-virus in 2011;31:2402–2409. the basilar artery lesions. Stroke 1986;17:1024–1028. 23. Lange C, Storkebaum E, de Almodovar CR, Dewerchin M, 6. Gilden D, White T, Khmeleva N, Boyer PJ, Nagel MA. Carmeliet P. Vascular endothelial growth factor: a neurovas- VZV in biopsy-positive and -negative giant cell arteritis: cular target in neurological diseases. Nat Rev Neurol 2016; analysis of 1001 temporal arteries. Neurol Neuroimmu- 12:439–454. nol Neuroinflamm 2016;3:e216. doi: 10.1212/NXI. 24. Zhang ZG, Zhang L, Jiang Q, et al. VEGF enhances 0000000000000216. angiogenesis and promotes blood-brain barrier leakage in 7. Gilden D, White T, Khmeleva N, et al. Prevalence and the ischemic brain. J Clin Invest 2000;106:829–838. distribution of VZV in temporal arteries of patients with 25. van Bruggen N, Thibodeaux H, Palmer JT, et al. VEGF giant cell arteritis. Neurology 2015;84:1948–1955. antagonism reduces edema formation and tissue damage 8. Nagel MA, Traktinskiy I, Azarkh Y, et al. Varicella zoster after ischemia/reperfusion injury in the mouse brain. virus vasculopathy: analysis of virus-infected arteries. Neu- J Clin Invest 1999;104:1613–1620. rology 2011;77:364–370. 26. Mellergard P, Sjogren F, Hillman J. Release of VEGF and 9. Nagel MA, Traktinskiy I, Stenmark KR, Frid MG, FGF in the extracellular space following severe subarach- Choe A, Gilden D. Varicella-zoster virus vasculopathy: noidal haemorrhage or traumatic head injury in humans. immune characteristics of virus-infected arteries. Neurology Br J Neurosurg 2010;24:261–267. 2013;80:62–68. 27. Slevin M, Krupinski J, Slowik A, Kumar P, Szczudlik A, 10. Sprague AH, Khalil RA. Inflammatory cytokines in vascu- Gaffney J. Serial measurement of vascular endothelial lar dysfunction and vascular disease. Biochem Pharmacol growth factor and transforming growth factor-beta1 in 2009;78:539–552. serum of patients with acute ischemic stroke. Stroke 11. Stenmark KR, Yeager ME, El Kasmi KC, et al. The adven- 2000;31:1863–1870. titia: essential regulator of vascular wall structure and func- 28. Pikula A, Beiser AS, Chen TC, et al. Serum brain-derived tion. Annu Rev Physiol 2013;75:23–47. neurotrophic factor and vascular endothelial growth factor 12. Dal Canto AJ, Swanson PE, O’Guin AK, Speck SH, Vir- levels are associated with risk of stroke and vascular brain gin HW. IFN-gamma action in the media of the great injury: Framingham Study. Stroke 2013;44:2768–2775. elastic arteries, a novel immunoprivileged site. J Clin 29. Arnold N, Girke T, Sureshchandra S, Nguyen C, Rais M, Invest 2001;107:R15–R22. Messaoudi I. Genomic and functional analysis of the host 13. Grose C, Brunel PA. Varicella-zoster virus: isolation and response to acute simian varicella infection in the lung. Sci propagation in human melanoma cells at 36 and 32 de- Rep 2016;6:34164. grees C. Infect Immun 1979;19:199–203. 30. Xu J, Lamouille S, Derynck R. TGF-beta-induced epithe- 14. Cohrs RJ, Wischer J, Essman C, Gilden DH. Character- lial to mesenchymal transition. Cell Res 2009;19: ization of varicella-zoster virus gene 21 and 29 proteins in 156–172. infected cells. J Virol 2002;76:7228–7238. 31. Jones D, Blackmon A, Neff CP, et al. Varicella-zoster virus 15. Zhang B, Shan H, Li D, et al. Different methods of de- downregulates programmed death ligand 1 and major his- taching adherent cells significantly affect the detection of tocompatibility complex class I in human brain vascular TRAIL receptors. Tumori 2012;98:800–803. adventitial fibroblasts, perineurial cells, and lung fibro- 16. Baggiolini M, Walz A, Kunkel SL. Neutrophil-activating blasts. J Virol 2016;90:10527–10534. peptide-1/interleukin 8, a novel cytokine that activates 32. Franz JK, Kolb SA, Hummel KM, et al. Interleukin-16, neutrophils. J Clin Invest 1989;84:1045–1049. produced by synovial fibroblasts, mediates chemoattrac- 17. Jones D, Alvarez E, Selva S, Gilden D, Nagel MA. Proin- tion for CD41 T lymphocytes in rheumatoid arthritis. flammatory cytokines and matrix metalloproteinases in Eur J Immunol 1998;28:2661–2671. CSF of patients with VZV vasculopathy. Neurol Neuro- 33. Liu MT, Chen BP, Oertel P, et al. The T cell chemo- immunol Neuroinflamm 2016;3:e246. doi: 10.1212/NXI. attractant IFN-inducible protein 10 is essential in host 0000000000000246. defense against viral-induced neurologic disease. 18. Haug A, Mahalingam R, Cohrs RJ, Schmid DS, Corboy J Immunol 2000;165:2327–2330. JR, Gilden D. Recurrent polymorphonuclear pleocytosis 34. Shen FH, Wang SW, Yeh TM, Tung YY, Hsu SM, Chen with increased red blood cells caused by varicella zoster SH. Absence of CXCL10 aggravates herpes stromal kera- virus infection of the central nervous system: case reports titis with reduced primary neutrophil influx in mice. and review of the literature. J Neurol Sci 2010;292:85–88. J Virol 2013;87:8502–8510. 19. Nadkami S, Dalli J, Hollywood J, Mason JC, Dasgupta B, 35. Schluns KS, Williams K, Ma A, Zheng XX, Lefrancois L. Perretti M. Investigational analysis reveals a potential role Cutting edge: requirement for IL-15 in the generation of

Neurology: Neuroimmunology & Neuroinflammation 9 primary and memory antigen-specific CD8 T cells. 38. Martinez-Taboada VM, Alvarez L, RuizSoto M, Marin- J Immunol 2002;168:4827–4831. Vidalled MJ, Lopez-Hoyos M. Giant cell arteritis and poly- 36. Iwasaki S, Minamisawa S, Yokoyama U, et al. Interleukin- myalgia rheumatica: role of cytokines in the pathogenesis and 15 inhibits smooth muscle cell proliferation and hyalur- implications for treatment. Cytokine 2008;44:207–220. onan production in rat ductus arteriosus. Pediatr Res 39. Ly KH, Regent A, Tamby MC, Mouthon L. Pathogenesis 2007;62:392–398. of giant cell arteritis: more than just an inflammatory con- 37. Cercek M, Matsumoto M, Li H, et al. Autocrine role of dition? Autoimmun Rev 2010;9:635–645. vascular IL-15 in intimal thickening. Biochem Biophys 40. Guillevin L, Regent A. Treating giant-cell arteritis: is IL-6 Res Commun 2006;339:618–623. the cytokine to target? Lancet 2016;387:1882–1883.

10 Neurology: Neuroimmunology & Neuroinflammation MRI evaluation of thalamic volume differentiates MS from common mimics

Andrew J. Solomon, MD ABSTRACT Richard Watts, PhD Objective: To determine whether MRI evaluation of thalamic volume differentiates MS from other Blake E. Dewey, BA disorders that cause MRI white matter abnormalities. Daniel S. Reich, MD, Methods: There were 40 study participants: 10 participants with MS without additional comor- PhD bidities for white matter abnormalities (MS 2 c); 10 participants with MS with additional comor- bidities for white matter abnormalities (MS 1 c); 10 participants with migraine, MRI white matter 2 Correspondence to abnormalities, and no additional comorbidities for white matter abnormalities (Mig c); and 10 Dr. Solomon: participants previously incorrectly diagnosed with MS (Misdx). T1-magnetization-prepared rapid [email protected] gradient-echo and T2-weighted three-dimensional fluid attenuation inversion recovery sequen- ces were acquired on a Phillips Achieva d-Stream 3T MRI, and scans were randomly ordered and de-identified for a blinded reviewer who performed MRI segmentation using LesionTOADS. Results: Mean normalized thalamic volume differed among the 4 cohorts (analysis of variance, p 5 0.005) and was smaller in the 20 MS participants compared with the 20 non-MS participants (p , 0.001), smaller in MS 2 c compared with Mig 2 c(p 5 0.03), and smaller in MS 1 ccompared with Misdx (p 5 0.006). The sensitivity and specificity were both 0.75 for diagnosis of MS with a thalamic volume ,0.0077. Conclusions: MRI volumetric evaluation of the thalamus, but not other deep gray-matter struc- tures, differentiated MS from other diseases that cause white matter abnormalities and are often mistaken for MS. Evaluation for thalamic atrophy may improve accuracy for diagnosis of MS as an adjunct to additional radiologic criteria. Thalamic volumetric assessment by MRI in larger cohorts of patients undergoing evaluation for MS is needed, along with the development of automated and easily applied volumetric assessment tools for future clinical application. Classification of evidence: This study provides Class III evidence that MRI evaluation of thalamic volume differentiates MS from other diseases that cause white matter abnormalities. Neurol Neuroimmunol Neuroinflamm 2017;4:e387; doi: 10.1212/NXI.0000000000000387

GLOSSARY 3D FLAIR 5 three-dimensional fluid attenuation inversion recovery; ANOVA 5 analysis of variance; CVS 5 central vein sign; DMT 5 disease modifying therapy; ROC 5 receiver operating characteristic; SVID 5 small vessel ischemic disease; T1- MPRAGE 5 T1-magnetization-prepared rapid gradient-echo.

The diagnosis of MS relies on an interpretation of clinical and radiographic data.1 The absence of a highly specific biomarker for MS in the setting of a considerable number of diseases and syndromes that can mimic its clinical and radiographic appearance2,3 makes accurate diagnosis challenging. Misdiagnosis of MS remains a problem with significant consequences for patients and health care systems.4–6 Patients incorrectly diagnosed with MS are exposed to unnecessary risks associated with disease modifying therapy (DMT) for MS, as well as resulting morbidities.6 Novel imaging techniques that facilitate differentiation of MS from other disorders may improve

From the Department of Neurological Sciences (A.J.S.) and Department of Radiology (R.W.), University of Vermont College of Medicine, Burlington; Department of Electrical and Computer Engineering (B.E.D.), Johns Hopkins University; and Translational Neuroradiology Section (B.E.D., D.S.R.), Division of Neuroimmunology and Neurovirology, National Institute of Neurological Disorders and Stroke, Bethesda, MD. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by University of Vermont, Department of Neurological Sciences. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 diagnostic accuracy. Thalamic atrophy has concatenation of the necessary transform matrices. Skull stripping 15 been identified early in the disease course of was performed using SPECTRE. Whole-brain and lesion segmentation was performed using 7 MS, including in pediatric MS, presymptom- LesionTOADS.16 LesionTOADS is a fully automated segmen- atic MS or “radiologically isolated syndrome,”8 tation algorithm that uses multichannel MRI (T1-MPRAGE and clinically isolated syndrome.9–11 Thus, and T2-FLAIR) data to simultaneously segment T2-lesions MRI assessment of thalamic volume is a suit- and the whole brain into major structures. This is done by combining the information of statistical atlases of segmentation able candidate for evaluation as an imaging likelihoods and a whole-brain topological atlas to create a topo- biomarker to distinguish MS from other dis- logically consistent segmentation. Volumes were all normalized orders. Although atrophy of multiple gray- to the intracranial volume calculated from the SPECTRE brain mask. matter structures has been demonstrated in The Shapiro-Wilk test was used to evaluate results for nor- MS, in several recent studies, assessment of mality. An analysis of variance (ANOVA) was used to test for dif- thalamic volume seemed to best differentiate ference between the 4 cohorts with a normal distribution. MS from neuromyelitis spectrum disorder.12,13 Individual contrasts using the ANOVA error term were used to compare the combined MS and non-MS cohorts as well as the This study aimed to determine whether MRI- various pairwise combinations of the 10-participant individual based volumetric assessment of the thalamus cohorts. A receiver operating characteristic (ROC) curve was gen- can differentiate MS from migraine and other erated to demonstrate sensitivity and specificity of thalamic vol- diseases that cause MRI white matter abnor- ume for MS diagnosis. Visual evaluation of the ROC curve was used to choose a cutoff value that might provide optimal sensitiv- malities and may mimic MS. ity and specificity for a diagnosis of MS. Sensitivity was defined as the probability of a thalamic volume less than the cutoff for pa- METHODS The primary research question is to determine tients with MS. Specificity was defined as the probability of a tha- whether MRI evaluation of thalamic volume differentiated MS lamic volume greater than or equal to the cutoff for non-MS from other disorders that cause MRI white matter abnormalities. patients. The methodology for this pilot study would provide Class III evi- dence for this research question. Standard protocol approvals, registrations, and patient Forty participants, comprising 4 cohorts, participated in the consents. The study was approved by the University of Vermont study. These included (1) 10 participants with a diagnosis of Institutional Review Board, and all assessments were performed at MS by 2010 criteria1 who also had no history of a comorbidity the University of Vermont. Written informed consent was ob- that may also cause MRI white matter abnormalities (MS 2 c); tained from all participants. (2) 10 participants with a diagnosis of MS by 2010 criteria1 with at least 1 additional comorbidity known to cause MRI white RESULTS Patient characteristics. Baseline demo- matter abnormalities (MS 1 c); (3) 10 participants with a diag- graphics of MS 2 c, MS 1 c, Mig 2 c, and Misdx nosis of migraine and a history of an MRI with at least 2 white cohorts are presented in table 1. There was no signif- matter abnormalities in any location but with no history of addi- icant difference among the 4 cohorts in age (ANOVA tional comorbidities known to cause white matter abnormalities 5 2 p 0.24). (Mig c); and (4) 10 participants who had been determined to 2 have been previously incorrectly diagnosed with MS, did not In the MS c cohort, 8/10 participants were meet 2010 criteria,1 and in whom a variety of diagnoses had been receiving DMT for MS at the time of participation identified to explain clinical and radiographic abnormalities mis- in the study. The 2 remaining participants were both taken for MS (Misdx). Diagnoses in the MS 2 c and MS 1 c 63 years old; one had never received DMT after diag- cohorts had been established after evaluation by a neurologist nosis in the early 1990s and the second discontinued 2 with MS subspecialty training. Diagnoses in the Mig c cohort DMT, because of questionable continued benefits, 6 had been established after evaluation by a neurologist. Diagnoses in the Misdx cohort were determined by a single neurologist with months before participating in the study. In the MS 1 MS subspecialty training at the University of Vermont during the c cohort, 9/10 participants were receiving DMT at course of a new patient evaluation, after a complete evaluation of the time of the study and 1 participant had received clinical history, neurologic examination, laboratory testing, and a final dose of alemtuzumab approximately 3 years neuroimaging was complete. In these patients, 2010 MS diagnos- before participating in the study. 1 tic criteria were not met, and alternative diagnoses were The comorbid conditions known to cause MRI identified. white matter abnormalities in the MS 1 c cohort T1-magnetization-prepared rapid gradient-echo (MPRAGE) and T2-weighted three-dimensional fluid attenuation inversion included migraine (4), hypertension (3), hyperten- recovery (3D FLAIR) sequences were acquired on a Philips Achie- sion and migraine (1), diabetes mellitus and migraine va d-Stream 3T MRI, and scans were randomly ordered and (1), and diabetes mellitus and hypertension (1). In de-identified for a blinded reviewer who performed MRI segmen- addition, 6/10 participants in this cohort had a history tation using LesionTOADS. Two registration steps were per- of comorbid tobacco use. formed using the Advanced Normalization Tools registration Mean duration of misdiagnosis in the Misdx framework.14 Each MPRAGE image was rigidly aligned to the MNI152 atlas, and the corresponding FLAIR scan was registered cohort was 9 years (median: 6 years). In this cohort, to the MPRAGE image, also using rigid alignment. All trans- 5/10 participants had received DMT for MS in the formations were performed using 1 interpolation step by past. CSF evaluation, including testing for elevation

2 Neurology: Neuroimmunology & Neuroinflammation Table 1 Study participant characteristics Figure 1 LesionTOADS thalamic segmentation depicted in one participant from each MS 2 c(n5 10) cohort

Age 44 (16)

Sex 9 F/1 M

Years since clinical onset of MS 9 (7)

Phenotype 10 RRMS

MS 1 c(n5 10)

Age 43 (9)

Sex 9 F/1 M

Years since clinical onset of MS 9 (6)

Phenotype 10 RRMS

Mig 2 c(n5 10)

Age 47 (13)

Sex 10 F

Misdx (n 5 10)

Age 53 (7)

Sex 9 F/1 M

Abbreviations: Mig 2 c 5 migraine with MRI white matter abnormalities without additional comorbidities for MRI white matter abnormalities; Misdx 5 previously misdiag- (A) MS without comorbidities for MRI white matter abnor- nosed with MS; MS 2 c 5 MS without comorbidities for malities, (B) MS with an additional comorbidity for MRI white MRI white matter abnormalities; MS 1 c 5 MS with addi- matter abnormalities, (C) migraine with MRI white matter tional comorbidities for MRI white matter abnormalities; abnormalities and without additional comorbidities for MRI RRMS 5 relapsing-remitting MS. white matter abnormalities, (D) previously misdiagnosed Values for age and years since clinical onset of MS are with MS. Dark red: CSF, dark orange: cortical gray matter, given as mean (SD). light orange: thalamus and striatum, white: white matter, and red: lesions. in intrathecal IgG or oligoclonal bands, was normal in the20MS(MS2 candMS1 c) compared with the 9 participants and had not been performed in 1 par- 20 non-MS (Mig 2 c and Misdx) cohorts (p , 0.001). ticipant. MRI of the cervical and thoracic spinal cord Comparing individual cohorts, mean thalamic had been performed in 6/10 participants and was nor- volume was smaller in MS 2 c compared with Mig mal; only cervical spinal cord MRI had been per- 2 c(p 5 0.03), in MS 1 c compared with Misdx formed in the remaining 4/10 participants and was (p 5 0.006), and in MS 1 c compared with Mig 2 c normal in all cases. The clinical diagnoses in this (p 5 0.002). There was no difference in mean tha- cohort included migraine (2), psychogenic disorder lamic volumes between MS 2 c and Misdx (p 5 and migraine (4), migraine and trigeminal neuralgia 0.09), MS 2 c and MS 1 c(p 5 0.24), or between (1), migraine and vitamin B12 deficiency (1), vertigo Mig 2 c and Misdx (p 5 0.63) cohorts. (1), and transient numbness (1). Diagnoses presumed responsible for abnormal brain MRI findings in this cohort included migraine (2), migraine and small ves- Table 2 Thalamic volume as proportion of sel ischemic disease (SVID) due to hypertension (1), intracranial volume migraine and SVID due to tobacco use (3), migraine and SVID due to tobacco use and hypertension (1), Cohort (n 5 10/cohort) Mean 6 SD SVID due to tobacco use (2), and migraine, vitamin MS 2 c 0.00755 6 0.00066 B12 deficiency, and SVID due to tobacco use (1). MS 1 c 0.00715 6 0.00090 6 Thalamic volumetric assessment. Figure 1 depicts seg- Mig 2 c 0.00830 0.00065 mentation performed by LesionTOADS in 1 partic- Misdx 0.00814 6 0.00075 ipant from each of the 4 cohorts. Normalized Abbreviations: Mig 2 c 5 migraine with MRI white matter thalamic volume differed among the 4 cohorts abnormalities and without additional comorbidities for MRI (ANOVA, p 5 0.005). The mean thalamic volumes white matter abnormalities; Misdx 5 previously misdiag- 2 5 of each cohort are presented in table 2, and figure 2 nosed with MS; MS c MS without comorbidities for MRI white matter abnormalities; MS 1 c 5 MS with addi- plots the mean thalamic volume of each participant in tional comorbidities for MRI white matter abnormalities. the 4 cohorts. Thalamic mean volumes were smaller in p value from analysis of variance: p 5 0.005.

Neurology: Neuroimmunology & Neuroinflammation 3 or putamen (p 5 0.08) mean volumes were different Figure 2 Thalamic volume normalized to intracranial volume for each 5 participant in the 4 cohorts between the MS and non-MS cohorts (p 0.88).

DISCUSSION In this pilot study, MRI evaluation of normalized thalamic volume differentiated partici- pants with MS from participants with other common causes of white matter abnormalities. The presence of diminished thalamic volume early in the course of MS compared with healthy controls,7–11 as well as the correlation between thalamic atrophy and clinical disability in MS, have been established over the last decade. Our findings are comparable to recent data12,13 reporting that thalamic volume was the gray-matter measure that best distinguished MS from neuromyelitis optica spectrum disorder. Thalamic demyelinating lesions,17 destructive conse- 2 1 MS c: MS without comorbidities for MRI white matter abnormalities, MS c: MS with 18 additional comorbidities for MRI white matter abnormalities, Mig 2 c: migraine with MRI quences of inflammation, and disruption of thalamo- 19 white matter abnormalities and without additional comorbidities for MRI white matter abnor- cortical connections are potential contributors to malities, Misdx: previously misdiagnosed with MS. thalamic atrophy in MS. Our data suggest that any disruption of thalamocortical tracts by white matter le- The sensitivity and specificity were both 0.75 for sions in other disorders is not associated with the same diagnosis of MS with normalized thalamic volume degree of thalamic volume loss seen in patients with MS. , 0.0077. MS participants were 3 times more likely The pathologic processes in migraine and SVID occur- , to have a thalamic volume 0.0077 compared with ring at the site of white matter lesions are distinct from participants in the non-MS cohort. Figure 3 shows MS, and these processes, as well as direct thalamic injury the ROC curve demonstrating sensitivity and speci- in MS,20 may be responsible for the volumetric differ- ficity of thalamic volume for MS diagnosis. ences we observed. The presence of cortical lesions in 21 Volumetric assessment of other gray-matter structures. MS that may also disrupt thalamocortical tract integ- There was no difference in caudate or putamen vol- rity and the absence of such lesions in migraine and 13,22 ume across the 4 cohorts (ANOVA, p 5 0.34 and other disorders that may mimic MS may have also p 5 0.09, respectively). Neither caudate (p 5 0.88) influenced differences in thalamic volume demonstrated between the MS and non-MS cohorts. Most participants (18/20) in our non-MS cohort had migraine and MRI white matter abnormalities. Figure 3 Receiver operating characteristic (ROC) curve demonstrating sensitivity and specificity of thalamic, caudate, and putamen volume Our findings are particularly interesting in this popu- for MS diagnosis lation,23,24 since thalamocortical network dysfunc- tion,25,26 thalamic microstructural changes,27,28 and thalamic morphological abnormalities29 have been recently demonstrated in individuals with migraine. Migraine is a disorder that is frequently clinically and radiographically mistaken for MS,5,6 and previous studies have also suggested an increased risk of migraine in patients with MS.30 In our MS 1 c cohort, 6/10 participants had migraine. The patho- physiology of white matter abnormalities in migraine,31 and the temporal progression of these lesions,32 differ from MS and could account for the differences in thalamic volume we observed. Recent studies have identified group-level gray-matter imag- ing abnormalities in migraine predominantly in the frontal and temporal lobes,33 and an association with lesions in the periaquductal gray matter and increased inflammatory activity in MS patients with migraine,34,35 suggesting the potential of MRI meth- ods for the differentiation of patients with and with- out migraine.

4 Neurology: Neuroimmunology & Neuroinflammation There were several limitations to this study. Partic- differentiated MS from other diseases that cause white ipants were recruited from a convenience sample. matter abnormalities and are often mistaken for MS. Although MS and migraine are more common in Recent data support the use of such volumetric assess- women, fewer men participated in the study. A pro- ment of regional atrophy in distinguishing diseases spective cohort evaluated at the time of clinical pre- with similar radiographic and phenotypical presenta- sentation would be a more suitable population to tion.13,40 Thalamic volumetric assessment by MRI in determine the sensitivity and specificity of MRI eval- larger cohorts of patients undergoing evaluation for uation of thalamic volume for a diagnosis of MS. MS is needed, along with development of automated Although migraine may be frequently mistaken for and easily applied volumetric assessment tools,13 for MS,4–6,36 participants in the Mig 2 c cohort were future clinical application. not suspected to have MS. The final diagnoses attrib- uted to clinical symptoms, and MRI abnormalities in AUTHOR CONTRIBUTIONS Andrew J. Solomon contributed to conceptualization and study design, the Misdx cohort lacked highly specific biomarkers, analysis and interpretation of the data, and drafting of the manuscript and it remains possible that specific diagnoses in some for intellectual content. Richard Watts and Blake E. Dewey contributed participants (though not the ruling out of MS) were to analysis and interpretation of the data and drafting of the manuscript incorrect. The Misdx cohort also did not comprise for intellectual content. Daniel S. Reich contributed to conceptualization and study design, analysis and interpretation of the data, and drafting of the full breadth of disorders that may be potentially the manuscript for intellectual content. mistaken for MS. Last, the small sample size of the MS 1 c and MS 2 c cohorts may have limited the ACKNOWLEDGMENT ability to detect differences in thalamic volumes asso- The authors thank Diantha Howard, MA, MS, Informatics Core Man- ager, University of Vermont Clinical Research Center, for data analysis; ciated with the presence of an additional comorbidity Jay Gonyea, MS, and Scott Hipko, BS, the University of Vermont MRI for white matter lesions in participants with MS. Center for Biomedical Imaging, for image acquisition. Although our study demonstrated a difference in normalized mean thalamic volumes between the STUDY FUNDING MS and non-MS cohorts, the overlap between these Supported by the University of Vermont Department of Neurological Sciences, University of Vermont Department of Radiology, the Univer- cohorts suggests that thalamic volumetric assessment sity of Vermont MRI Center for Biomedical Imaging, and the Intramural alone is unlikely to distinguish MS from other disor- Research Program of NINDS. ders. However, further data may demonstrate that a very low thalamic volume aids the confirmation of DISCLOSURE a diagnosis of MS. Furthermore, evaluation for tha- A.J. Solomon consulted for Biogen, served on the scientific advisory board for Teva, Genentech, and EMD Serono. R. Watts received lamic atrophy may show promise as an adjunct to research support from NSF and NIH. B.E. Dewey reports no disclosures. additional MS radiologic criteria. Incorporation of D.S. Reich received research support from Vertex Pharmaceuticals, thalamic volumetric assessment into current MS diag- NINDS, and Myelin Repair Foundation. Go to Neurology.org/nn for full disclosure forms. nostic criteria1 could in principle demonstrate improved specificity and sensitivity for MS; however, Received March 16, 2017. Accepted in final form June 13, 2017. whether this would be the case in the setting of con- sistent and proper application of current radiographic REFERENCES criteria—which is not always straightforward4,6—is 1. Polman CH, Reingold SC, Banwell B, et al. Diagnostic unknown. Instead, evaluation of thalamic volume criteria for multiple sclerosis: 2010 revisions to the McDo- – might be incorporated into algorithms currently in nald criteria. Ann Neurol 2011;69:292 302. 2. Brownlee WJ, Hardy TA, Fazekas F, Miller DH. Diagno- development using new imaging techniques for MS sis of multiple sclerosis: progress and challenges. Lancet diagnosis. For instance, the recently developed 2017;389:1336–1346. FLAIR* MRI sequence allows evaluation of MS 3. Toledano M, Weinshenker BG, Solomon AJ. A clinical white matter lesions for a central vein—the “central approach to the differential diagnosis of multiple sclerosis. vein” sign (CVS)—and differentiates MS from other Curr Neurol Neurosci Rep 2015;15:57. populations.37,38 The sensitivity of MRI for visualiza- 4. Solomon AJ, Weinshenker BG. Misdiagnosis of multiple sclerosis: frequency, causes, effects, and prevention. Curr tion of MS cortical lesions presents methodological Neurol Neurosci Rep 2013;13:403. challenges; yet, the detection of such cortical lesions 5. Solomon AJ, Klein EP, Bourdette D. “Undiagnosing” may also improve specificity of current imaging cri- multiple sclerosis: the challenge of misdiagnosis in MS. teria for MS.39 Thus, an imaging algorithm that Neurology 2012;78:1986–1991. might incorporate assessment of CVS, detection of 6. Solomon AJ, Bourdette DN, Cross AH, et al. The cortical lesions, and evaluation of thalamic volume contemporary spectrum of multiple sclerosis misdiag- nosis: a multicenter study. Neurology 2016;87:1393– may demonstrate improved sensitivity and specificity 1399. for MS compared with current imaging criteria. 7. Aubert-Broche B, Fonov V, Ghassemi R, et al. Regional In this study, MRI evaluation of thalamic vol- brain atrophy in children with multiple sclerosis. Neuro- umes, but not other deep gray-matter structures, image 2011;58:409–415.

Neurology: Neuroimmunology & Neuroinflammation 5 8. Azevedo CJ, Overton E, Khadka S, et al. Early CNS neu- 25. Coppola G, Di Renzo A, Tinelli E, et al. Thalamo-cortical rodegeneration in radiologically isolated syndrome. Neurol network activity between migraine attacks: insights from Neuroimmunol Neuroinflamm 2015;2:e102. doi: 10. MRI-based microstructural and functional resting-state 1212/NXI.0000000000000102. network correlation analysis. J Headache Pain 2016;17: 9. Bergsland N, Horakova D, Dwyer MG, et al. Subcortical 100. and cortical gray matter atrophy in a large sample of pa- 26. Coppola G, Di Renzo A, Tinelli E, et al. Thalamo-cortical tients with clinically isolated syndrome and early relapsing- network activity during spontaneous migraine attacks. remitting multiple sclerosis. AJNR Am J Neuroradiol Neurology 2016;87:2154–2160. 2012;33:1573–1578. 27. Coppola G, Tinelli E, Lepre C, et al. Dynamic changes in 10. Henry RG, Shieh M, Okuda DT, Evangelista A, Gorno- thalamic microstructure of migraine without aura patients: Tempini ML, Pelletier D. Regional grey matter atrophy in a diffusion tensor magnetic resonance imaging study. Eur J clinically isolated syndromes at presentation. J Neurol Neurol 2014;21:287–e13. Neurosurg Psychiatry 2008;79:1236–1244. 28. Granziera C, Daducci A, Romascano D, et al. Structural 11. Zivadinov R, Bergsland N, Dolezal O, et al. Evolution of abnormalities in the thalamus of migraineurs with aura: cortical and thalamus atrophy and disability progression in a multiparametric study at 3 T. Hum Brain Mapp 2014; early relapsing-remitting MS during 5 years. AJNR Am J 35:1461–1468. Neuroradiol 2013;34:1931–1939. 29. Magon S, May A, Stankewitz A, et al. Morphological 12. Hyun JW, Park G, Kwak K, et al. Deep gray matter abnormalities of thalamic Subnuclei in migraine: a multi- atrophy in neuromyelitis optica spectrum disorder and center MRI study at 3 Tesla. J Neurosci 2015;35:13800– multiple sclerosis. Eur J Neurol 2017;24:437–445. 13806. 13. Eshaghi A, Wottschel V, Cortese R, et al. Gray matter 30. Gelfand AA, Gelfand JM, Goadsby PJ. Migraine and mul- MRI differentiates neuromyelitis optica from multiple tiple sclerosis: epidemiology and approach to treatment. sclerosis using random forest. Neurology 2016;87:2463– Mult Scler Relat Disord 2013;2:73–79. 2470. 31. Takano T, Tian GF, Peng W, et al. Cortical spreading 14. Avants BB, Tustison NJ, Stauffer M, Song G, Wu B, Gee depression causes and coincides with tissue hypoxia. Nat JC. The Insight ToolKit image registration framework. Neurosci 2007;10:754–762. Front Neuroinform 2014;8:44. 32. Hamedani AG, Rose KM, Peterlin BL, et al. Migraine and 15. Carass A, Cuzzocreo J, Wheeler MB, Bazin PL, Resnick white matter hyperintensities: the ARIC MRI study. Neu- SM, Prince JL. Simple paradigm for extra-cerebral tissue rology 2013;81:1308–1313. removal: algorithm and analysis. Neuroimage 2011;56: 33. Rocca MA, Ceccarelli A, Falini A, et al. Brain gray matter 1982–1992. changes in migraine patients with T2-visible lesions: a 3-T 16. Shiee N, Bazin PL, Ozturk A, Reich DS, Calabresi PA, Pham MRI study. Stroke 2006;37:1765–1770. DL. A topology-preserving approach to the segmentation of 34. Gee JR, Chang J, Dublin AB, Vijayan N. The association of brain images with multiple sclerosis lesions. Neuroimage brainstem lesions with migraine-like headache: an imaging 2010;49:1524–1535. study of multiple sclerosis. Headache 2005;45:670–677. 17. Vercellino M, Plano F, Votta B, Mutani R, Giordana MT, 35. Graziano E, Hagemeier J, Weinstock-Guttman B, Ramas- Cavalla P. Grey matter pathology in multiple sclerosis. amy DP, Zivadinov R. Increased contrast enhancing lesion J Neuropathol Exp Neurol 2005;64:1101–1107. activity in relapsing-remitting multiple sclerosis migraine 18. Herranz E, Gianni C, Louapre C, et al. Neuroinflamma- patients. Neuroimage Clin 2015;9:110–116. tory component of gray matter pathology in multiple scle- 36. Liu S, Kullnat J, Bourdette D, et al. Prevalence of brain rosis. Ann Neurol 2016;80:776–790. magnetic resonance imaging meeting Barkhof and McDo- 19. Henry RG, Shieh M, Amirbekian B, Chung S, Okuda nald criteria for dissemination in space among headache DT, Pelletier D. Connecting white matter injury and tha- patients. Mult Scler 2013;19:1101–1105. lamic atrophy in clinically isolated syndromes. J Neurol Sci 37. Sati P, Oh J, Constable RT, et al. The central vein sign 2009;282:61–66. and its clinical evaluation for the diagnosis of multiple 20. Harrison DM, Oh J, Roy S, et al. Thalamic lesions in sclerosis: a consensus statement from the North American multiple sclerosis by 7T MRI: clinical implications and Imaging in Multiple Sclerosis Cooperative. Nat Rev Neu- relationship to cortical pathology. Mult Scler 2015;21: rol 2016;12:714–722. 1139–1150. 38. Solomon AJ, Schindler MK, Howard DB, et al. “Central 21. Calabrese M, Filippi M, Gallo P. Cortical lesions in mul- vessel sign” on 3T FLAIR* MRI for the differentiation of tiple sclerosis. Nat Rev Neurol 2010;6:438–444. multiple sclerosis from migraine. Ann Clin Transl Neurol 22. Absinta M, Rocca MA, Colombo B, et al. Patients with 2016;3:82–87. migraine do not have MRI-visible cortical lesions. J Neurol 39. Preziosa P, Rocca M, Mesaros S, et al. Comparison of 2012;259:2695–2698. MRI criteria for the diagnosis of multiple sclerosis: role 23. Kruit MC, van Buchem MA, Hofman PA, et al. Migraine of cortical lesions (S45.003). Neurology 2016:86(16 as a risk factor for subclinical brain lesions. JAMA 2004; suppl). 291:427–434. 40. Guevara C, Bulatova K, Barker GJ, Gonzalez G, Crossley 24. Aradi M, Schwarcz A, Perlaki G, et al. Quantitative MRI NA, Kempton MJ. Whole-brain atrophy differences studies of chronic brain white matter hyperintensities in between progressive Supranuclear Palsy and Idiopathic migraine patients. Headache 2013;53:752–763. Parkinson’s disease. Front Aging Neurosci 2016;8:218.

6 Neurology: Neuroimmunology & Neuroinflammation a4-integrin receptor desaturation and disease activity return after natalizumab cessation

Tobias Derfuss, MD* ABSTRACT * John M. Kovarik, PhD Objective: To describe the time course of a4-integrin receptor desaturation and disease activity Ludwig Kappos, MD return in patients with relapsing-remitting MS who discontinued natalizumab and to investigate Marina Savelieva, PhD baseline and on-study predictors for the recurrence of disease activity. Richa Chhabra, Methods: In the course of TOFINGO, a 32-week, patient- and rater-blinded multicenter, parallel- MS (Pharm) group study, we performed MRI, counted relapses, and measured a4-integrin receptor occupancy Avinash Thakur, (RO) at baseline and 8, 12, 16, 20, and 24 weeks. The relationship between RO and total number MS (Pharm) of new T1 gadolinium-enhancing (Gd1) lesions was modeled using Poisson linear regression. Ying Zhang, PhD Heinz Wiendl, MD Results: Patients (N 5 142) were randomized (1:1:1) to 8-, 12-, or 16-week washout (WO) groups. Davorka Tomic, DVM At randomization, the median RO in the 8-, 12-, and 16-week WO groups was 94.5%, 92.4%, and 90.9%, which declined to 79.8%, 30.7%, and 8.7% after 8, 12, and 16 weeks of WO, respectively. The percentage of patients with new T1 Gd1 lesions increased with longer WO Correspondence to period before commencing fingolimod: 2.1% (8 weeks), 9.1% (12 weeks), and 50.0% (16 weeks). Dr. Derfuss: Overall, 71% of patients with first relapse between weeks 6 and 18 had RO values below the [email protected] time-matched population median. Higher T2 lesion volume (LV) at baseline predicted a higher number of new T1 Gd1 lesions. Conclusions: A faster decline in natalizumab RO, longer WO period, and higher T2 LV at baseline were associated with an increased risk for return of inflammatory disease activity. These results provide a mechanistic rationale and, together with the main outcomes of the TOFINGO study, support initiation of fingolimod within 8 weeks of natalizumab discontinuation. ClinicalTrials.gov identifier: NCT01499667. Neurol Neuroimmunol Neuroinflamm 2017;4:e388; doi: 10.1212/NXI.0000000000000388

GLOSSARY FACS 5 fluorescence-assisted cell sorter; Gd1 5 gadolinium enhancing; LNI 5 last natalizumab infusion; LV 5 lesion volume; RO 5 receptor occupancy; RRMS 5 relapsing-remitting MS; WO 5 washout.

Patients with relapsing-remitting MS (RRMS) often discontinue natalizumab treatment, most frequently due to the risk of developing progressive multifocal leukoencephalopathy.1 Natalizumab discontinuation leads to recurrence of clinical and radiologic disease activity within 4 months of discontinuation.2 Natalizumab primarily acts via blocking the a4- integrin receptor and preventing potentially autoaggressive immune cell transition through the blood-brain barrier.2 Hence, a4-integrin receptor desaturation should be a direct mea- sure of its therapeutic activity. No systematic studies have reported the relationship between a4-integrin receptor occupancy (RO) and recurrence of disease activity following natalizu- mab discontinuation. Supplemental data at Neurology.org/nn *These authors contributed equally to the manuscript. From the Neurological Clinic and Policlinic (T.D.), Departments of Medicine and Biomedicine, University Hospital Basel; Novartis Pharma AG (J.M.K., M.S., D.T.), Basel; Neurological Clinic and Policlinic (L.K.), Departments of Medicine, Clinical Research, Biomedicine and Biomedical Engineering, University Hospital Basel, Switzerland; Novartis Pharmaceuticals Corporation (Y.Z.), East Hanover, NJ; and Department of Neurology (H.W.), University of Münster, Germany; Novartis Healthcare Pvt. Ltd. (R.C.. A.T.), Hyderabad, India. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by Novartis. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 natalizumab WO groups before fingolimod therapy was initiated: Figure 1 8-week WO (no treatment for 8 weeks since LNI followed by 24 weeks of treatment with oral fingolimod 0.5 mg); 12-week WO (no treatment for 8 weeks since LNI and placebo for 4 weeks followed by 20 weeks of treatment with fingolimod); or 16-week WO (no treatment for 8 weeks since LNI and placebo for 8 weeks followed by 16 weeks of treatment with fingolimod). All patients in the study population provided written informed consent. TOFINGO involved 44 study centers and was conducted in accordance with the ethical principles of the Declaration of Helsinki.

Study assessments. Natalizumab serum concentration and a4-integrin RO assessment. Patient visits were scheduled at screening, randomization, and weeks 8, 12, 16, 20, 24, and 32. At each visit, the serum concentration of natalizumab was deter- mined using a validated ELISA method (lower quantification limit, 0.1 mg/mL). Venous blood samples were collected at all Receptor desaturation after the last natalizumab dose. Median a4-integrin receptor occu- visits to determine a4-integrin RO (%) using a fluorescence- pancy by the WO group at screening (24 weeks), randomization (week 0), and during drug assisted cell sorter (FACS) analysis. Cells were stained with WO from weeks 8 to 32. Bars represent the IQR. A total of 137 patients (96.5%) provided matched isotype control antibodies to determine the specificity of at least 1 evaluable receptor occupancy measurement. IQR 5 interquartile range; WO 5 a4 integrin and natalizumab staining. The signal obtained with washout. matched isotype controls was used to set the threshold for CD49d-specific a4-integrin staining and/or antinatalizumab staining with anti-IgG4 reagent. Natalizumab (final concentra- In daily practice, fingolimod 0.5 mg is fre- tion, 10 mg) was spiked in 150 mL whole blood samples. The quently considered a suitable option for treat- samples were incubated with detection antibodies, CD49d- ment of patients with RRMS after specific a4-integrin staining, CD3 staining, and/or anti- natalizumab staining with anti-IgG4 reagent, for 15 minutes in natalizumab discontinuation, but the optimal the dark at 23°C. BD FACS lysing solution (2 mL) was added to 3 washout (WO) period remains unknown. the sample and incubated for 15 minutes in the dark at 23°C. TOFINGO, a randomized double-blind The samples were then pelleted by centrifugation and washed study, addressed the relationship between the twice with 2 mL of FACS Flow Sheath Fluid. One hundred thousand nucleated events were then analyzed on FACS Calibur, duration of natalizumab WO before fingoli- using FSC/CD3 staining for T-lymphocyte gating and determi- mod initiation and recurrence of disease activ- nation of a4-integrin and natalizumab-positive CD3 cells and ity.4 The results of TOFINGO suggested that median fluorescence intensity of a4-integrin and natalizumab staining. better control of disease activity is achieved if Assessment of disease recurrence. MRI assessments (T1 fingolimod is initiated within 8–12 weeks after gadolinium-enhancing [Gd1] lesion counts; new T2 lesions) natalizumab discontinuation, compared with were conducted at baseline and at weeks 8, 12, 16, 20, and 24. All 16 weeks.4 In the course of TOFINGO, relapses during the study (both confirmed and not confirmed) were counted. blood samples were collected to determine The association between RO and freedom from T1 Gd1 a4-integrin RO. We investigated the time lesions was explored using median-effect analysis.5 The relation- course of a4-integrin receptor desaturation ship between RO, patient-specific covariates, and total number 1 and its temporal relation to MRI and clinical of T1 Gd lesions was modeled using Poisson linear regression (e-Methods at Neurology.org/nn). disease activity. Other possible baseline and on-study predictors for the recurrence of dis- RESULTS Patient disposition and baseline characteristics. ease activity were also assessed. The analysis Overall, 142 patients were randomized (8-week was aimed to provide a mechanistic rationale WO: n 5 50; 12-week WO: n 5 42; and 16-week for the timing of switch from natalizumab to WO: n 5 50) and analyzed according to the WO fingolimod in patients with RRMS. group assigned at randomization. As reported pre- viously, baseline demographic and MS characteristics 4 METHODS Study design and patients. TOFINGO were well balanced across the WO groups. Of the (ClinicalTrials.gov identifier: NCT01499667), a 32-week, 142 patients, 137 (96.5%) provided $1 evaluable patient- and rater-blinded, randomized, multicenter, parallel- RO measurement. The percentage of patients pro- group study, included 142 patients with RRMS who received viding evaluable RO values at each visit averaged their last natalizumab infusion (LNI; 300 mg IV) within 67 days 72% (range, 58%–83%). The median serum con- of randomization. Details of the study have been published pre- viously.4 Briefly, patients with RRMS aged 18–65 years who had centration of natalizumab at the end of WO for the received natalizumab treatment for $6 months immediately 8-, 12-, and 16-week WO groups was 0 mg/mL, with before screening were randomized 1:1:1 to one of the following 3 a range of 0–1.07, 0–0.383, and 0–0.308 mg/mL,

2 Neurology: Neuroimmunology & Neuroinflammation RO and first report of T1 Gd1 lesions or first relapse. At Figure 2 randomization, the median RO was 94.5%, 92.4%, and 90.9% in the 8-, 12-, and 16-week WO groups, respectively. Overall, the median RO declined to 79.8%, 30.7%, and 8.7% at weeks 8, 12, and 16, respectively (figure 1). MRI scans were available for 121 patients. As the length of the WO period before fin- golimod initiation increased, the percentage of patients free from T1 Gd1 lesions decreased, although the relationship did not reach statistical significance (median-effect analysis, r2 5 0.891, p 5 0.214; figure 2A). Up to week 8, the median RO was relatively high (79.4%), and the incidence of new T1 Gd1 lesions was 2.1% (n 5 2). Both these patients with new T1 Gd1 lesions had ROs below the population median. By week 12, the median RO had declined to 17.9%, and the percentage of patients with new T1 Gd1 lesions was 9.1%. At week 16, the median RO was 9.35%, and the percentage of patients with new T1 Gd1 lesions was the highest (50.0%; figure 2B). Similar results were observed with RO and the first reporting of new/ enlarging T2 lesions (data not shown). Overall, 31 pa- tients had at least 1 relapse, 26 of whom had RO data available. Of these, 17 had the first relapse between weeks 6 and 18. The majority (n 5 12; 71%) of these 17 patients had RO values below the time-matched population median (figure e-1).

Model-based prediction of T1 Gd1 lesions and new/ (A) Median-effect modeling of receptor occupancy and freedom from T1 Gd1 lesions. enlarging T2 lesions. High T2 lesion volume (T2 LV) Scatter plot of the percentage of patients free from T1 Gd1 lesions vs median a4-integrin at baseline was associated with a higher number of T1 receptor occupancy for the 8-, 12-, and 16-week WO groups. Only patients not Gd1 lesions after LNI. As shown in figure 3, patients yet on fingolimod were included in the analysis. The curve represents the fit of the , median-effect model to the data. (B) Median receptor occupancy and first reporting of with baseline T2 LV 75th percentile experienced T1 Gd1 lesions. Scatter plot showing the a4-integrin receptor occupancy at the visit fewer new T1 Gd1 lesions than did those with a base- when first T1 Gd1 lesions were reported in the context of the population median receptor line T2 LV . 75th percentile. This was observed in 15 5 desaturation curve. Gd gadolinium enhancing; WO washout. all WO groups, most noticeable in the 16-week WO group (e-Results). respectively (table e-1), indicating that in most pa- tients, the drug was eliminated from blood in the DISCUSSION Patients with RRMS frequently switch first 2 months after LNI. to fingolimod after natalizumab discontinuation to

Figure 3

Total number of T1 Gd1 lesions when the baseline volume of T2 lesions was less than (A) and greater than (B) the 75th quartile. Bottom of box is Q1 and the top of the box is Q3; upper whiskers are 1.5 3 IQR of the hinge; lower whiskers are 1.5 3 IQR of the hinge. Gd15gadolinium enhancing; IQR 5 interquartile range.

Neurology: Neuroimmunology & Neuroinflammation 3 maintain adequate disease control.6,7 Several observa- of the study, analysis and interpretation of the data, and drafting and revis- tional studies and the randomized rater- and patient- ing the manuscript. R. Chhabra; drafting and revising the manuscript. A. Thakur: drafting and revising the manuscript. blinded TOFINGO study have suggested that a natalizumab WO period of .12 weeks is associated ACKNOWLEDGMENT with higher risk of disease reactivation,4,6 and that The authors thank the patients from whom data were taken for analysis. a WO period of ,12 weeks is associated with lower They thank Rebecca Gottschalk (Novartis Pharmaceuticals Corpora- tion, East Hanover, NJ) for coordination of the study. All authors edited MRI activity without being associated with increased the manuscript for intellectual content, provided guidance during 4,6,8 risks. Our study also suggests that in 80%̴ of manuscript development, and approved the final version submitted patients, disease recurrence (new T1 Gd1 lesions) is for publication. low up to week 12 during the WO period before fingolimod initiation. Significant disease return is STUDY FUNDING The study was sponsored by Novartis Pharma AG. observed between weeks 12 and 16 of the WO period , at an RO of 20% (figure 2A). This might indicate DISCLOSURE that an RO lower than 90% is still sufficient to suppress T.DerfussservesonscientificadvisoryboardsforNovartisPharmaceut- disease activity. However, recent small studies have icals, Merck Serono, Biogen, Sanofi Genzyme, GeNeuro, Octapharma, shown that some patients may relapse or have recur- MedDay, Mitsubishi Pharma, Roche, and Bayer Schering Pharma; has received funding for travel and/or speaker honoraria from Biogen, Sanofi rence of MRI activity within 8 weeks of LNI, indi- Genzyme, Novartis Pharmaceuticals, Merck Serono, Roche, and Bayer 9,10 cating possibilities of an early switch to fingolimod. Schering Pharma; is on the editorial board for PLoS One; receives research While a long WO period between LNI and fingolimod support from Biogen, Novartis Pharma, the European Union, the Swiss initiation may increase the risk of disease recurrence, National Foundation, and the Swiss MS Society; is a member of steering committee for Mitsubishi Pharma and GeNeuro; and his spouse is an early initiation of fingolimod may result in additive employee of and owns stock options in Novartis Pharma. J.M. Kovarik is effects on protective immune functions. on the editorial board for Journal of Clinical Pharmacology and is em- Despite a small sample size, the TOFINGO study ployed by and holds stock in Novartis. L. Kappos’ institution (University Hospital Basel) has received in the last 3 years and used exclusively for showed distinct numerical trends in pharmacody- research support: steering committee, advisory board, and consultancy namics, imaging, and clinical outcomes, which can fees (Actelion, Addex, Bayer HealthCare, Biogen Idec, Biotica, Genzyme, help determine the risk of disease recurrence after na- Lilly, Merck, Mitsubishi, Novartis, Ono Pharma, Pfizer, Receptos, talizumab discontinuation in individual patients. Sanofi, Santhera, Siemens, Teva, UCB, and Xenoport); speaker fees (Bayer HealthCare, Biogen Idec, Merck, Novartis, Sanofi, and Teva); Our results showed that while natalizumab is support for educational activities (Bayer HealthCare, Biogen, CSL cleared from the serum within 8–12 weeks after Behring, Genzyme, Merck, Novartis, Sanofi, and Teva); license fees for discontinuation, it still occupies a4-integrin recep- Neurostatus products; and grants (Bayer HealthCare, Biogen Idec, tors, thus contributing to disease activity control. European Union, Merck, Novartis, Roche Research Foundation, Swiss MS Society, and Swiss National Research Foundation). L. Kappos is on In view of the association of declining RO with the editorial board for Multiple Sclerosis Journal, Multiple Sclerosis and recurring disease activity, RO may emerge as a more Related Disorders, and Journal of Neurology. M. Sevelieva is employed plausible measure for an individualized choice of the by Novartis Pharma. R. Chhabra is employed by Novartis Healthcare. A. Thakur is employed by Novartis Healthcare. Y. Zhang is employed by optimal WO period. Ideally, decisions regarding the Novartis Pharma. H. Wiendl served on the scientific advisory board for optimal WO period would also take the individual Bayer Healthcare, Biogen Idec, Sanofi Genzyme, Merck Serono, risk profile into account. In addition to prenatalizumab Novartis, Roche, and Teva; is on the editorial board for PLoS One, relapse rate and MRI activity, which have previously Neurotherapeutics, and Recent Patents on Inflammation and Allergy Drug Discovery; has received compensation for serving as a consultant or been identified as risk factors, our study suggests speaker for, or has received research support from Bayer Schering that a high baseline T2 LV defines a patient popula- Pharma, Biogen Idec/Elan Corporation, CSL Behring, EMD Serono, tion with an increased risk of recurrence after Fresenius Medical Care, OmniaMed, GlaxoSmithKline, GW Pharmaceut- discontinuation. icals,BayerVital,SanofiGenzyme,Roche,MerckSerono,Novartis,Novo Nordisk, Sanofi Aventis, Teva Pharmaceutical Industries, German Ministry for Education and Research, European Union, Interdisciplinary Centre of Clinical Research, and PML Consortium; and Else Kroner Fresenius Foun- AUTHOR CONTRIBUTIONS dation, Fresenius Foundation, Hertie Foundation, and RE Children’sFoun- The study was designed by the sponsor, Novartis Pharma AG, in collabo- dation. D. Tomic holds a patent on fingolimod’s activity on T1 hypointense ration with the steering committee. Data were collected by the investigators lesions; is employed by Novartis; and holds stock in Novartis. Go to and analyzed by the sponsor. Novartis contributed to the interpretation of Neurology.org/nn for full disclosure forms. the study. All authors had full access to the data and had final responsibility for the contents and decision to submit for publication. T. Derfuss and Received May 15, 2017. Accepted in final form June 27, 2017. J.M. Kovarik: design and conceptualization of the study, analysis and inter- pretation of the data, and drafting and revising the manuscript. L. Kappos: REFERENCES analysis and interpretation of the data and revising the manuscript. 1. Bloomgren G, Richman S, Hotermans C, et al. Risk of M. Savelieva: design and conceptualization of the study, statistical analysis natalizumab-associated progressive multifocal leukoence- and interpretation of the data, and revising the manuscript. H. Wiendl: – design and conceptualization of the study, analysis and interpretation of phalopathy. N Engl J Med 2012;366:1870 1880. the data, and drafting and revising the manuscript. Y. Zhang: design and 2. Rasenack M, Derfuss T. Disease activity return after na- conceptualization of the study, statistical analysis and interpretation of the talizumab cessation in multiple sclerosis. Expert Rev Neu- data, and revising the manuscript. D. Tomic: design and conceptualization rother 2016;16:587–594.

4 Neurology: Neuroimmunology & Neuroinflammation 3. Klotz L, Grutzke B, Eveslage M, et al. Assessment of immune closely related to duration of wash out and alpha-4 integrin functions and MRI disease activity in relapsing-remitting desaturation—a prospective study in RRMS. Mult Scler J multiple sclerosis patients switching from natalizumab to 2015;21(11 suppl):556–557. fingolimod (ToFingo-Successor). BMC Neurol 2015;15:96. 8. Cohen M, Maillart E, Tourbah A, et al. Switching from 4. Kappos L, Radue EW, Comi G, et al. Switching from natalizumab to fingolimod in multiple sclerosis: a French natalizumab to fingolimod: a randomized, placebo- prospective study. JAMA Neurol 2014;71:436–441. controlled study in RRMS. Neurology 2015;85:29–39. 9. Fox RJ, Cree BA, De Seze J, et al. MS disease activity in 5. Chou TC, Talalay P. Quantitative analysis of dose-effect RESTORE: a randomized 24-week natalizumab treatment relationships: the combined effects of multiple drugs or interruption study. Neurology 2014;82:1491–1498. enzyme inhibitors. Adv Enzyme Regul 1984;22:27–55. 10. Naegelin Y, Rasenack M, Sanderson N, et al. An observa- 6. Jokubaitis VG, Li V, Kalincik T, et al. Fingolimod after tional study evaluating disease control, safety, and immu- natalizumab and the risk of short-term relapse. Neurology nological changes in patients with relapsing remitting 2014;82:1204–1211. multiple sclerosis switching from previous treatment with 7. Naegelin Y, Rasenack M, Blatti C, et al. Switching from natalizumab to fingolimod (SWITCH-UHBS). Mult Scler natalizumab to fingolimod: recurrence of disease activity is J 2013;19(11 Suppl):460.

Neurology: Neuroimmunology & Neuroinflammation 5 Teriflunomide slows BVL in relapsing MS A reanalysis of the TEMSO MRI data set using SIENA

Ernst-Wilhelm Radue, ABSTRACT * MD Objective: To assess, using structural image evaluation using normalization of atrophy (SIENA), * Till Sprenger, MD the effect of teriflunomide, a once-daily oral immunomodulator, on brain volume loss (BVL) in pa- Laura Gaetano, PhD tients with relapsing forms of MS enrolled in the phase 3 TEMSO study. Nicole Mueller-Lenke, Methods: TEMSO MR scans were analyzed (study personnel masked to treatment allocation) MD using SIENA to assess brain volume changes between baseline and years 1 and 2 in patients trea- Steve Cavalier, MD ted with placebo or teriflunomide. Treatment group comparisons were made via rank analysis of Karthinathan Thangavelu, covariance. PhD Michael A. Panzara, MD Results: Data from 969 patient MRI visits were included in this analysis: 808 patients had base- Jessica E. Donaldson, line and year 1 MRI; 709 patients had baseline and year 2 MRI. Median percentage BVL from PhD baseline to year 1 and year 2 for placebo was 0.61% and 1.29%, respectively, and for terifluno- Fiona M. Woodward, mide 14 mg, 0.39% and 0.90%, respectively. BVL was lower for teriflunomide 14 mg vs placebo PhD at year 1 (36.9% relative reduction, p 5 0.0001) and year 2 (30.6% relative reduction, p 5 Jens Wuerfel, MD 0.0001). Teriflunomide 7 mg was also associated with significant reduction in BVL vs placebo Jerry S. Wolinsky, MD over the 2-year study. The significant effects of teriflunomide 14 mg on BVL were observed in Ludwig Kappos, MD both patients with and without on-study disability worsening. Conclusions: The significant reduction of BVL vs placebo over 2 years achieved with teriflunomide is consistent with its effects on delaying disability worsening and suggests a neuroprotective Correspondence to potential. Dr. Sprenger: [email protected] Classification of evidence: Class II evidence shows that teriflunomide treatment significantly re- or Dr. Wuerfel: duces BVL over 2 years vs placebo. [email protected] ClinicalTrials.gov identifier: NCT00134563. Neurol Neuroimmunol Neuroinflamm 2017;4:e390; doi: 10.1212/NXI.0000000000000390

GLOSSARY BPF 5 brain parenchymal fraction; BVL 5 brain volume loss; CDW 5 confirmed disability worsening; DMT 5 disease- modifying therapy; EAE 5 experimental autoimmune encephalitis; EDSS 5 Expanded Disability Status Scale; MRIAP 5 MRI analysis package; SIENA 5 structural image evaluation using normalization of atrophy; SIENAX 5 SIENA cross- sectional method.

Accelerated brain volume loss (BVL) starts in the early stages of MS and is associated with accu- mulating physical and cognitive disability.1–5 In a recent meta-analysis of clinical trials including patients with relapsing-remitting MS, treatment effects of disease-modifying therapies (DMTs) on BVL showed a strong correlation with treatment effects on disability worsening over 2 years.6 In 2 phase 3 pivotal trials (TEMSO, NCT00134563, and TOWER, NCT00751881), teri- flunomide 14 mg (AUBAGIO, Genzyme, Cambridge, MA), a once-daily oral immunomodu- lator approved for relapsing-remitting MS,7–9 significantly reduced the risk of 12-week Supplemental data at Neurology.org/nn *These authors contributed equally to the analysis. From the Medical Image Analysis Center (MIAC AG) (E.-W.R., L.G., N.M.-L., J.W.), Basel, Switzerland; DKD HELIOS Klinik (T.S.), Wiesbaden, Germany; Neurologic Clinic and Policlinic (T.S., L.G., L.K.), University Hospital Basel and University of Basel, Switzerland; Sanofi Genzyme (S.C., K.T.), Previously Sanofi Genzyme (M.A.P.), and WAVE Life Sciences (M.A.P.), Cambridge, MA; Fishawack Communications Ltd (J.E.D., F.M.W.), Abingdon, Oxfordshire, UK; and McGovern Medical School (J.S.W.), UTHealth, Houston, TX. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by Sanofi Genzyme. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 confirmed disability worsening (CDW) in pa- Neurological Institute Z coordinates 210 to 160 mm), which tients with relapsing MS.10,11 A significant was selected for optimal reproducibility and comparability to pre- vious trials of other oral DMTs.15–21 MR images of 2 time points reduction in annualized relapse rate in both were coregistered, and surface changes were determined using the studies and a reduction in MRI activity in fully automated algorithm to estimate brain volume changes (fig- TEMSO were also reported (no MRI was per- ure e-1 at Neurology.org/nn). formed in TOWER).10–12 BVL was assessed in In addition, the single-point SIENA cross-sectional method (SIENAX)13 was also applied to the T1-weighted images to esti- TEMSO by changes in brain parenchymal mate the normalized whole-brain volume of each patient at base- fraction (BPF) using an MRI analysis package line (figure e-2). (MRIAP), developed and performed at Study personnel were masked to treatment allocation and other study data. A rigorous quality control of all uploaded MRI data and UTHealth, Houston, TX. The results demon- all stages of the SIENA analysis was performed. This led to the strated numerical reduction of BVL by ;25% exclusion of some scans for 1 or more of the following reasons: with teriflunomide 14 mg vs placebo, but this required MRI sequence(s) for the evaluation were missing, quality was not statistically significant.12 The MRIAP of MRI sequence(s) was insufficient for the evaluation, and refer- ence or longitudinal follow-up MRI was missing or excluded from method has had limited use in successful large the analysis for one of the aforementioned reasons. phase 3 studies in MS; therefore, it is difficult Statistical analysis. The change from baseline in annualized to assess its sensitivity for detecting change and percentage brain volume was analyzed, and median (95% CI) val- therapeutic effects across different DMTs. ues were determined for each treatment group. Treatment group Structural image evaluation using normaliza- comparisons were made via nonparametric analysis of covariance, adjusted for region, age, Expanded Disability Status Scale (EDSS) tion of atrophy (SIENA) is a well-established, strata, and normalized brain volume (using SIENAX). A similar longitudinal, registration-based technique method was applied for treatment group comparisons in a sub- with a low error rate in detecting brain volume group analysis of BVL in patients with or without on-study changes over time.13,14 CDW, defined as an increase from baseline of at least 1.0 point in the EDSS score (or at least 0.5 points for patients with Given the consistent and statistically signif- a baseline EDSS score greater than 5.5) that persisted for at least icant effects of teriflunomide on reducing the 12 or 24 weeks. The relationship between CDW and BVL was risk of disability worsening, and the strong evaluated using a Cox model with CDW as the dependent vari- able and BVL as the independent variable, together with other association between BVL and disability wors- covariates (treatment and stratification factors used for random- ening in MS, a new masked analysis of ization). As this SIENA analysis was a masked reanalysis, but not TEMSO MRI was undertaken using SIENA. a prospectively defined analysis as per the TEMSO protocol, no formal adjustment for multiplicity was performed. METHODS Standard protocol approvals, registrations, Classification of evidence. This study provides Class II evi- and patient consents. The TEMSO study (ClinicalTrials.gov dence of reduction in BVL, as assessed by masked SIENA analy- identifier NCT00134563) was conducted in accordance with the sis, over 2 years of treatment with once-daily oral teriflunomide International Conference on Harmonisation Guidelines for Good 7 mg or 14 mg, compared with placebo, in patients with relapsing Clinical Practice and the Declaration of Helsinki. The protocol forms of MS. was approved by central and local ethics committees and the respective institutional review board; patients provided written RESULTS Patients. At each time point, $89.5% of informed consent before entering the study. the patient scans included in the original MRI anal- Study design. TEMSO was a 2-year, phase 3, randomized, ysis of the TEMSO study (described as the TEMSO double-blind, placebo-controlled, parallel-group study designed to MRI population below) were deemed suitable for evaluate the efficacy and safety of teriflunomide in reducing the fre- inclusion at the corresponding time point in the SIENA quency of relapses and progression of physical disability in patients analysis: 969 patients had baseline MRI; 808 had who had relapsing MS, as reported previously.10 After a screening phase of up to 4 weeks, eligible patients were randomly assigned (in baseline and year 1 MRI (scans from 95/903 patients a 1:1:1 ratio) to receive a once-daily oral dose of placebo, 7 mg of [10.5%] with evaluable MRI in the TEMSO MRI teriflunomide, or 14 mg of teriflunomide for 108 weeks. population, were excluded at year 1, before SIENA The primary objective of the SIENA reanalysis of the analysis); and 709 had baseline and year 2 MRI (scans TEMSO MRI data set was to evaluate the effect of teriflunomide from 69/778 patients [8.9%] were excluded at year 2). vs placebo on BVL from baseline to year 1 and year 2. The mean (SD) normalized brain volume values at MRI analysis. MR scans collected from the TEMSO study12 at baseline for those patients included in the SIENA baseline and at weeks 48 (year 1) and 108 (year 2) were analyzed analysis were similar between the 3 treatment groups: using the SIENA method, as performed at the Medical Image 1,508.9 cm3 (77.9 cm3), 1,507.8 cm3 (83.1 cm3), and Analysis Center (MIAC AG, Basel, Switzerland). Further details 1,502.7 cm3 (75.8 cm3), for placebo and terifluno- of the specific SIENA methodology used for this analysis are 5 provided in the supplementary material. mide 7 and 14 mg, respectively (p 0.5610; for the SIENA was applied to 3-mm thick precontrast T1-weighted comparison between groups of baseline normalized images of a 70-mm central brain area section (Montreal brain volume).

2 Neurology: Neuroimmunology & Neuroinflammation between the groups analyzed: those to be included Table Patient demographics and baseline clinical characteristicsa in the new SIENA analysis only, those included in

TEMSO MRI population SIENA analysis the original TEMSO MRI population only, and (N 5 903) (N 5 808) those excluded from the new SIENA analysis Age, mean (SD), y 37.9 (8.7) 37.8 (8.7) (figure e-3). Female, n (%) 653 (72.3) 602 (74.5) Relationship between on-study disability worsening and Time since first diagnosis, mean (SD), y 5.29 (5.42) 5.26 (5.51) BVL. In a Cox model, BVL over 2 years had an effect Number of relapses, mean (SD) on CDW: risk reduction of 11% for every 1% change

Within past 1 year 1.4 (0.7) 1.4 (0.7) in brain volume (p 5 0.0203).

Within past 2 years 2.2 (1.0) 2.2 (1.0) The normalized brain volume at baseline was lower in the patient subgroup with subsequent on- Baseline EDSS score, mean (SD) 2.61 (1.29) 2.59 (1.29) study disability worsening vs the subgroup without MS treatment in previous 2 years, n (%) 236 (26.1) 208 (25.7) disability worsening. At baseline, patients with and Baseline brain parenchymal fraction,b 0.76 (0.024) 0.76 (0.024) mean (SD) without 12-week CDW had mean (SD) brain vol- umes of 1,495 cm3 (83.4 cm3) and 1,509 cm3 Abbreviations: EDSS 5 Expanded Disability Status Scale; SIENA 5 structural image eval- (77.6 cm3), respectively; p 5 0.02. Patients with uation using normalization of atrophy. a Demographics shown for patients with a baseline and year 1 scan; similar data were seen and without 24-week CDW had brain volumes of 3 3 3 3 for patients with a baseline and year 2 scan, as well as patients with MRI of sufficient 1,493 cm (86.7 cm ) and 1,509 cm (77.5 cm ), quality at all time points. respectively; p 5 0.03. b From original TEMSO analysis. Placebo-treated patients demonstrated increased rates of BVL if they also had on-study disability wors- ening vs those without worsening (figure 2A). Of Baseline demographics and clinical characteristics note, BVL remained statistically significantly lower of patients included in the SIENA analysis were similar in patients treated with teriflunomide 14 mg than to those of patients in the previously published in placebo-treated patients, regardless of the presence TEMSO MRI population (table), including between- or absence of on-study disability worsening at both treatment comparisons (data not shown). Furthermore, years 1 and 2 (figure 2, B and C). The median per- the annualized relapse rate was consistent between both centage change from baseline in patients treated with populations with respect to the comparison between teriflunomide 14 mg with vs without on-study placebo-treated patients (TEMSO MRI population: 12-week CDW at year 1 was 20.25 vs 20.40 (p 5 0.517; SIENA: 0.509) and between patients treated 0.48), respectively, and at year 2 was 20.90 vs 20.87 with teriflunomide 14 mg (TEMSO MRI population: (p 5 0.68), respectively. 0.343; SIENA: 0.330). The percentage of patients with A similar pattern of reduced BVL was observed in CDW was also comparable within the placebo-treated patients receiving teriflunomide 7 mg, although re- patients (TEMSO MRI population: 25.2%; SIENA: sults did not reach statistical significance in the sub- 23.6%) and the teriflunomide 14 mg–treated patients group of patients without 12- or 24-week CDW (TEMSO MRI population: 18.8%; SIENA: 19.8%) from baseline to year 2 (figure e-4, A and B). in both study populations. These data demonstrate that the exclusion of a small number of scans, due to DISCUSSION Limiting BVL in MS is increasingly stringent quality control as done for the SIENA recognized as an evolving and important therapeutic analysis, did not bias the final outcomes. goal, despite challenges of measurement at an individ- SIENA analysis. At both time points analyzed, teriflu- ual patient level. The importance of BVL is due, in nomide 7 and 14 mg significantly slowed BVL com- part, to the recognized association between BVL pared with placebo (figure 1A). The difference in and long-term accumulation of physical and cogni- BVL with teriflunomide at year 1 (vs placebo) was tive disability, which reduce both patients’ functional maintained at year 2. Analysis of BVL in patients with ability and their quality of life. Accelerated BVL can MRI for all 3 time points (i.e., baseline, year 1, and be evident from the earliest stages of MS, which year 2) demonstrated a similar pattern (figure 1B). highlights the need for early and effective interven- In a sensitivity analysis, exclusion of patients who tion.1,22,23 In this masked reanalysis of the TEMSO had been included in the original TEMSO MRI pop- MRI data set using the SIENA method, teri- ulation (assessed using MRIAP) but who did not have flunomide significantly slowed BVL vs placebo, an MRI data of sufficient quality for the SIENA analysis effect that was maintained to the end of the TEMSO did not affect the overall findings. This was demon- core study. strated by similar patterns of brain volume changes SIENA is a well-established longitudinal registration- (as assessed by changes in BPF using MRIAP) based technique with a low error rate in detecting brain

Neurology: Neuroimmunology & Neuroinflammation 3 Figure 1 Annualized percentage brain volume change

Annualized median percentage change from baseline in BVL over 2 years for teriflunomide 14 and 7 mg vs placebo: (A) using all scans available at each time point; (B) for patients with complete scan series at all time points. BVL 5 brain volume loss; CI 5 confidence interval; *Relative change vs placebo.

volume changes over time13,14 and has been repeat- in this reanalysis. It is acknowledged that exclusion of edly used to measure treatment effects on brain atro- scans as a result of rigorous quality-control has the phy with oral DMTs other than teriflunomide.15–21 potential to introduce selection bias. Nonetheless, SIENA can outperform many cross-sectional meth- a sensitivity analysis demonstrated that exclusion of ods regarding sensitivity and reproducibility13,24 and, some TEMSO MRI did not affect baseline or other thus, may be better suited for this purpose.12 characteristics of the patients assessed. This analysis of TEMSO MRI using the SIENA The significant reduction in BVL observed in this methodology was conducted using the same proce- analysis is consistent with the positive effect of teriflu- dures as in other phase 3 studies of oral DMTs,15–21 nomide on disability worsening already initially with study personnel masked to treatment allocation observed in a phase 2 study and later confirmed in and other study data. By applying rigorous quality- both phase 3 studies: TEMSO and TOWER, which control criteria, some TEMSO MRIs were excluded compared teriflunomide with placebo in patients with

4 Neurology: Neuroimmunology & Neuroinflammation Figure 2 BVL according to on-study disability worsening

BVL according to on-study disability worsening in (A) placebo-treated patients with and without 12- or 24-week confirmed disability worsening (CDW); (B) placebo- and teriflunomide 14 mg–treated patients with 12- or 24-week CDW; (C) placebo- and teriflunomide 14 mg–treated patients without 12- or 24-week CDW. BVL 5 brain volume loss; CI 5 confidence interval. *Relative change vs patients without CDW; **Relative change vs placebo. relapsing forms of MS.10,11,25 Our additional analysis of disability worsening. On-study disability worsen- showing faster rates of BVL in placebo-treated pa- ing may also reflect the clinical manifestations of tients with on-study disability worsening compared accumulated BVL before the start of the TEMSO with those without worsening provides further evi- study, as evidenced by lower baseline brain volumes dence that greater BVL is related to an increased risk in the subgroup with disability worsening vs the

Neurology: Neuroimmunology & Neuroinflammation 5 subgroup without. Nevertheless, treatment with teri- conceptualization of the study and/or analysis or interpretation of the flunomide was associated with a significant reduction data. L. Gaetano and K. Thangavelu: analysis or interpretation of the data. E.-W. Radue, T. Sprenger, L. Gaetano, N. Mueller-Lenke, in BVL, regardless of the presence of on-study dis- S. Cavalier, K. Thangavelu, M.A. Panzara, J.E. Donaldson, F.M. ability worsening, suggesting that teriflunomide ex- Woodward, J. Wuerfel, J.S. Wolinsky, and L. Kappos: drafting or erts this effect in a broad range of patients with MS. revising the manuscript for intellectual content. In a meta-analysis of data from 13 randomized ACKNOWLEDGMENT clinical trials evaluating DMTs in patients with The authors acknowledge the contribution of the TEMSO study in- RRMS, the size of the treatment effect on brain vestigators, the members of UTHealth, Houston, TX, the central atrophy was closely correlated with the size of MRI center for the TEMSO Study, and Pascal Kuster and Stefan treatment effect on 2-year disability worsening at Traud of MIAC AG for their technical assistance and support in data preparation and processing pipeline handling. This manuscript was re- 2 5 6 the trial level (R 0.48). The meta-analysis viewedbyLarisaMiller,PharmD,ofSanofiGenzyme.Draftingand included the data from the original MRIAP analy- editorial support for this manuscript were provided by Jessica Donald- sis of BPF changes conducted in TEMSO.12 When son and Fiona Woodward, of Fishawack Communications Ltd. the 13-trial meta-analysis was repeated, replacing (Abingdon, UK), and was funded by Sanofi Genzyme.

the MRIAP analysis of TEMSO MRI with the STUDY FUNDING 2 SIENA analysis of BVL, the R values for the treat- Supported by Sanofi Genzyme. ment effect correlation between brain atrophy and disability worsening strengthened from 0.48 to DISCLOSURE 0.61.26 E.-W. Radue received speaker honoraria and travel compensation from Preservation of brain volume with teriflunomide is Bayer Schering, Biogen, Fondazione Italiana Sclerosi Multipla, Genzyme, Novartis, Merck Serono, MorphoSys, and Synthon; consulted for Bayer consistent with its proposed mechanism of action in Schering, Biogen Idec, Fondazione Italiana Sclerosi Multipla, Genzyme, MS. The primary mechanism of action of terifluno- Novartis, Merck Serono, Synthon, and MorphoSys; and received institu- mide is believed to relate to the reversible inhibition tional research support from Novartis, Biogen Idec, Actelion, Basilea, SAKK, and Synarc. T. Sprenger served on the scientific advisory board of proliferation of activated T and B lymphocytes, for Sanofi Genzyme, Novartis, Actelion, Electrocore, and Mitsubishi thereby limiting their involvement in damaging pro- Pharma; received speaker honoraria from Biogen, Sanofi Genzyme, Teva, cesses within the CNS.27 Studies of teriflunomide Novartis, and Desitin; and received research support from Novartis administered to rats with experimental autoimmune Pharma, EFIC-Grunenthal, Swiss National Science Foundation, and Swiss MS Society. L. Gaetano served on the scientific advisory board encephalitis (EAE) at disease onset showed reduced for Novartis Pharma AG and interned for Novartis Pharma AG. N. lymphocyte counts in the rats’ spinal cords.28 EAE Mueller-Lenke reports no disclosures. S. Cavalier is employed by Sanofi animals dosed therapeutically with teriflunomide Genzyme. K. Thangavelu is employed by Sanofi Genzyme. M.A. Panzara showed reduced CNS inflammation, reduced disease was previously employed by Sanofi Genzyme and is currently employed by WAVE Life Sciences. J.E. Donaldson is employed by Fishawack Com- scores, reduced axonal damage and demyelination, municationsLtd.F.M.WoodwardisemployedbyFishawackCommunica- and preserved sensory and motor neuronal function tions Ltd. J. Wuerfel served on the scientific advisory board for Novartis, compared with untreated EAE animals.29,30 However, Biogen, Genzyme, Teva, and Roche; received travel support and/or speaker honoraria from Bayer, Biogen, and Novartis; is employed by MI- whether these results are driven solely by the AC AG; and received research support from the German Ministry of anti-inflammatory effects of the drug acting in the Education and Research and the German Ministry of Economy. J.S. Wo- periphery is currently unclear. Evidence is emerging linsky served on the scientific advisory board for AbbVie, Alkermes, Bayer that teriflunomide is found in the CNS at Healthcare, Biogen, Bionest, Celgene, Clene Nanomedicine, EMD Sero- no, Forward Pharma A/S, MedDay, Novartis, Roche/Genentech, Sanofi/ pharmacologically relevant concentrations and that Genzyme, Takeda, and Teva; received travel funding and/or speaker hon- teriflunomide treatment in vitro has direct effects oraria from Academic CME, ACTRIMS, CMSC, ECTRIMS, Masters on activated rodent microglia and astrocyte func- MS, Medscape, PRIME, and WebMD; served on the editorial board for Multiple Sclerosis Journal and Multiple Sclerosis and Related Diseases; tions,31 suggesting that teriflunomide may have holds a patent for Methods for treating CNS Lesions; consulted for potential neuroprotective effects within the CNS, at AbbVie, Alkermes, Bayer Healthcare, Biogen, Bionest, Celgene, Clene least in rodents. Nanomedicine, Forward Pharma A/S, EMD Serono, Novartis, Roche/ In this new analysis, teriflunomide—an immuno- Genentech, Sanofi/Genzyme, Misspelt, and Teva; received research support from NIH; and received royalties for monoclonal antibodies modulatory DMT with established efficacy on annu- out-licensed through the University of Texas Health Science Center at alized relapse rate, disability worsening, and MRI Houston to Millipore. L. Kappos served on the editorial board for lesion activity—has been demonstrated to reduce Multiple Sclerosis Journal, Multiple Sclerosis and Related Disorders, BVL. This result—besides indicating a neuroprotec- and Journal of Neurology; received research support from Actelion, — Alkermes, Almirall, Bayer, Biogen, Excemed, GeNeuro SA, tive potential of teriflunomide further strengthens Genzyme, Merck, Mitsubishi Pharma, Novartis, Receptos, Roche, the evidence of a link between BVL and disability Sanofi-Aventis, Santhera, Teva, Vianex, the Swiss MS Society, the worsening. Swiss National Research Foundation, the European Union, and Roche Research Foundations; and receives license fees for Neuro- status products paid directly to University Hospital Basel. Go to AUTHOR CONTRIBUTIONS Neurology.org/nn for full disclosure forms. E.-W. Radue, T. Sprenger, L. Gaetano, N. Mueller-Lenke, S. Cavalier, M.A. Panzara, J. Wuerfel, J.S. Wolinsky, and L. Kappos: design or Received March 3, 2017. Accepted in final form June 27, 2017.

6 Neurology: Neuroimmunology & Neuroinflammation REFERENCES remitting multiple sclerosis: results from the phase 3 1. De Stefano N, Airas L, Grigoriadis N, et al. Clinical rel- FREEDOMS II study. Presented at the 28th Congress of evance of brain volume measures in multiple sclerosis. the European Committee for Treatment and Research in CNS Drugs 2014;28:147–156. Multiple Sclerosis; October 10–13, 2012; Lyon, France. 2. Riley C, Azevedo C, Bailey M, Pelletier D. Clinical appli- 18. Arnold DL, Gold R, Kappos L, et al. Effects of delayed- cations of imaging disease burden in multiple sclerosis: release dimethyl fumarate on MRI measures in the Phase 3 MRI and advanced imaging techniques. Expert Rev DEFINE study. J Neurol 2014;261:1794–1802. Neurother 2012;12:323–333. 19. Gold R, Kappos L, Arnold DL, et al. Placebo-controlled 3. Popescu V, Agosta F, Hulst HE, et al. Brain atrophy and phase 3 study of oral BG-12 for relapsing multiple sclero- lesion load predict long term disability in multiple sclero- sis. N Engl J Med 2012;367:1098–1107. sis. J Neurol Neurosurg Psychiatry 2013;84:1082–1091. 20. Fox RJ, Miller DH, Phillips JT, et al. Placebo-controlled 4. Calabrese M, Agosta F, Rinaldi F, et al. Cortical lesions phase 3 study of oral BG-12 or glatiramer in multiple and atrophy associated with cognitive impairment in sclerosis. N Engl J Med 2012;367:1087–1097. relapsing-remitting multiple sclerosis. Arch Neurol 2009; 21. Miller DH, Fox RJ, Phillips JT, et al. Effects of delayed- 66:1144–1150. release dimethyl fumarate on MRI measures in the phase 3 5. Marcus JF, Waubant EL. Updates on clinically isolated CONFIRM study. Neurology 2015;84:1145–1152. syndrome and diagnostic criteria for multiple sclerosis. 22. Radue EW, Barkhof F, Kappos L, et al. Correlation Neurohospitalist 2013;3:65–80. between brain volume loss and clinical and MRI outcomes 6. Sormani MP, Arnold DL, De Stefano N. Treatment effect in multiple sclerosis. Neurology 2015;84:784–793. on brain atrophy correlates with treatment effect on dis- 23. Mowry EM, Beheshtian A, Waubant E, et al. Quality ability in multiple sclerosis. Ann Neurol 2014;75:43–49. of life in multiple sclerosis is associated with lesion 7. Genzyme. AUBAGIO (Summary of Product Characteris- burden and brain volume measures. Neurology 2009; tics). Available at: ema.europa.eu/docs/en_GB/ 72:1760–1765. document_library/EPAR_-_Product_Information/human/ 24. Durand-Dubief F, Belaroussi B, Armspach JP, et al. Reli- 002514/WC500148682.pdf. Accessed December 5, 2016. ability of longitudinal brain volume loss measurements 8. Genzyme. AUBAGIO (Package Insert). Available at: between 2 sites in patients with multiple sclerosis: com- accessdata.fda.gov/drugsatfda_docs/label/2016/202992s002 parison of 7 quantification techniques. AJNR Am J Neu- lbl.pdf. Accessed December 8, 2016. roradiol 2012;33:1918–1924. 9. Bar-Or A, Pachner A, Menguy-Vacheron F, Kaplan J, 25. Vidal-Jordana A, Sastre-Garriga J, Rovira A, Montalban X. Wiendl H. Teriflunomide and its mechanism of action Treating relapsing-remitting multiple sclerosis: therapy ef- in multiple sclerosis. Drugs 2014;74:659–674. fects on brain atrophy. J Neurol 2015;262:2617–2626. 10. O’Connor P, Wolinsky JS, Confavreux C, et al. Random- 26. Sormani MP, Radue EW, Sprenger T, et al. Incorporating ized trial of oral teriflunomide for relapsing multiple scle- the TEMSO SIENA analysis improves correlation of brain rosis. N Engl J Med 2011;365:1293–1303. atrophy and disability progression. Presented at the 2nd 11. Confavreux C, O’Connor P, Comi G, et al. Oral terifluno- Congress of the European Academy of Neurology; May mide for patients with relapsing multiple sclerosis 28-31, 2016; Copenhagen, Denmark. (TOWER): a randomised, double-blind, placebo-controlled, 27. Li L, Liu J, Delohery T, Zhang D, Arendt C, Jones C. The phase 3 trial. Lancet Neurol 2014;13:247–256. effects of teriflunomide on lymphocyte subpopulations in 12. Wolinsky JS, Narayana PA, Nelson F, et al. Magnetic human peripheral blood mononuclear cells in vitro. resonance imaging outcomes from a phase III trial of teri- J Neuroimmunol 2013;265:82–90. flunomide. Mult Scler 2013;19:1310–1319. 28. Ringheim GE, Lee L, Laws-Ricker L, et al. Teriflunomide 13. Smith SM, Zhang Y, Jenkinson M, et al. Accurate, robust, attenuates immunopathological changes in the dark agouti and automated longitudinal and cross-sectional brain rat model of experimental autoimmune encephalomyelitis. change analysis. Neuroimage 2002;17:479–489. Front Neurol 2013;4:169. 14. Smith SM, Zhang Y, Jenkinson M, et al. TR01SMS1: 29. Merrill JE, Hanak S, Pu SF, et al. Teriflunomide reduces Brain Atrophy Analysis Using Single- and Multiple- behavioral, electrophysiological, and histopathological Time-Point Data. FMRIB Technical Note. 2001. Avail- deficits in the dark Agouti rat model of experimental able at: fmrib.ox.ac.uk/datasets/techrep/tr01ss1/tr01ss1. autoimmune encephalomyelitis. J Neurol 2009;256: pdf. Accessed February 3, 2017. 89–103. 15. Kappos L, Radue EW, O’Connor P, et al. A placebo- 30. Iglesias-Bregna D, Hanak S, Ji Z, et al. Effects of pro- controlled trial of oral fingolimod in relapsing multiple phylactic and therapeutic teriflunomide in transcranial sclerosis. N Engl J Med 2010;362:387–401. magnetic stimulation–induced motor-evoked potentials 16. Calabresi PA, Radue EW, Goodin D, et al. Safety and in the dark agouti rat model of experimental autoim- efficacy of fingolimod in patients with relapsing-remitting mune encephalomyelitis. J Pharmacol Exp Ther 2013; multiple sclerosis (FREEDOMS II): a double-blind, rando- 347:203–211. mised, placebo-controlled, phase 3 trial. Lancet Neurol 31. Edling E, Woodworth L, Agrawal R, et al. Teriflunomide 2014;13:545–556. impacts primary microglia and astrocyte functions in vitro. 17. Radue EW, Goodin D, Jeffery D, et al. Fingolimod re- 32nd Congress of the European Committee for Treatment duces magnetic resonance imaging inflammatory lesion and Research in Multiple Sclerosis; September 14-17, activity versus placebo in patients with relapsing– 2016; London, UK.

Neurology: Neuroimmunology & Neuroinflammation 7 VIEWS & REVIEWS Frequencies of neuronal autoantibodies in healthy controls Estimation of disease specificity

Katharina Lang ABSTRACT Harald Prüss, MD Objective: To provide an extensive overview on the prevalence of antibodies against neuronal sur- faces (neuronal surface antibody [NSAb]) in healthy participants and disease controls.

Correspondence to Methods: We searched the PubMed database (1974 to October 2016) for studies that analyzed Dr. Prüss: frequencies of 22 different NSAbs in serum or CSF and included controls. Antibody prevalence [email protected] was calculated for patients with NSAb-mediated disease and controls, including healthy partici- pants, and those with neurologic and nonneurologic diseases. Different assays for antibody detection were compared. Results: In 309 articles, 743,299 antibody tests for 22 NSAbs were performed, including 30,485 tests for 19 NSAbs in healthy controls (HCs). Of these, 26,423 (86.7%) were tested with current standard methods, usually cell-based assays. Prevalence was very low in HCs (mean 0.23%, absent for 9/19 antibodies), and test numbers ranged from 21 to 3,065 per antibody. One study reported .1,000 healthy participants, and the others contained 21–274 samples. CSF samples were virtu- ally not available from HCs. NSAb prevalence was considerably higher (1.5%) in 69,850 disease controls, i.e., patients not initially suspected to have NSAb-mediated diseases. Antibody determina- tion in controls using nonstandard assays (such as ELISA) resulted in 6% positivity. Conclusions: NSAbs are rarely found in healthy participants, particularly with standard detection methods, suggesting high disease specificity and supporting their diagnostic usefulness. Conversely, positive titers in atypical patients might point to the still expanding phenotypic spectrum. Future stud- ies should include more CSF samples, data from HCs, and experimental evidence for antibody pathogenicity. Neurol Neuroimmunol Neuroinflamm 2017;4:e386; doi: 10.1212/NXI.0000000000000386

GLOSSARY AMPAR1/2, AMPAR3 5 alpha-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor subunits 1/2 and 3; AQP4 5 aquaporin-4; Caspr2 5 contactin-associated protein-like 2; CBA 5 cell-based assay; D2R 5 dopamine-2 receptor; DNER 5 delta/notch-like epidermal growth factor–related receptor; DPPX 5 dipeptidyl-peptidase-like protein 6; FACS 5 fluores- cence-activated cell sorting; GABAaR and GABAbR 5 gamma-aminobutyric acid-A and -B receptor; GlyR 5 glycine receptor; HC 5 healthy control; IgLON5 5 cell adhesion molecule IgLON family member 5; IHC 5 immunohohstochemistry; Lgi1 5 leucine-rich glioma-inactivated protein 1; mGluR5 5 metabotropic glutamate receptor 5; MOG 5 myelin oligodendrocyte glycoprotein; nAChR (m), nAChR (g) 5 nicotinic acetylcholine receptor of muscle (m) and ganglionic (g) type; NMDAR/NR1, NMDAR/NR2 5 N-methyl-D-aspartate-receptor subunits NR1 or NR2; NSAb 5 neuronal surface antibody; PRG5 5 plas- ticity-related gene 5; RIA 5 radioimmunoassay; VGKC 5 voltage-gated potassium channel; WB 5 Western blot.

Over the past years, Neurology and Psychiatry have witnessed a breathtaking development in the field of neuronal surface antibody (NSAb)-mediated diseases. Since the description of an encephalitis associated with antibodies against the NMDA receptor (NMDAR) in 2007,1 more than 15 new NSAbs have been identified as the pathogenic agents in diverse neuropsychiatric syndromes.2–7 The relative novelty of the antibodies and the associated, still expanding clinical pictures occasion- ally cause a dilemma. The presence of an atypical neuropsychiatric syndrome together with positive NSAb titers not always allows the conclusion that the antibody is causing the disease. The antibody might indeed be causative for a broader phenotype than initially appreciated or merely represent an Supplemental data at Neurology.org/nn From the German Center for Neurodegenerative Diseases (DZNE) Berlin (K.L., H.P.); and Department of Neurology and Experimental Neurology (K.L., H.P.), Charité—Universitätsmedizin Berlin, Germany. Funding information and disclosures are provided at the end of the article. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by German Center for Neurodegenerative Diseases (DZNE) Berlin. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 unspecific finding or a false-positive test result. antibody tests for the 22 most relevant NSAbs, In other words, when a test for NSAbs in clinical using human serum or CSF (figure 1 and table practice returns positive in blood or CSF, how e-1 at Neurology.org/nn). likely is an underlying antibody-mediated disease? METHODS Literature research. This review article com- prehensively analyzed the PubMed database literature from This question of NSAb specificity is highly 1974 to October 2016 using the search terms “antibody” and relevant since positive test results are the com- “encephalitis” together with the antigens: acetylcholine recep- mon rationale for an (often aggressive) immu- tor, nAChR, AMPA receptor, AQP4, aquaporin-4, caspr2, d2 notherapy. To date, there is no systematic receptor, dopamine-2 receptor, DNER, DPPX, GABAa receptor, GABAb receptor, glycine receptor, IgLON5, lgi1, literature review analyzing the frequency of metabotropic glutamate receptor 5, mGluR5, MOG, neuro- NSAbs in control populations. In particular, fascin, NMDAR, and VGKC.

knowledge of the frequency in healthy partic- Selection criteria. Articles were considered that tested for $1of ipants would be a valuable source to determine the 22 IgG antibodies and included (1) serum or CSF samples of disease specificity of autoantibodies, as they— a control population in addition to the disease group and (2) by definition—lack symptoms of an antibody- provided the exact number of tested control samples. For NSAbs with only few publications (D2R, DNER, DPPX, GlyR, mediated disease. The purpose of this study is IgLON5, mGluR5, PRG5, and neurexin-3a), we considered all to provide an extensive overview on NSAb published articles.

prevalence in healthy and disease controls, Analysis. The cohorts were categorized into 3 groups—based on evaluating data from more than 700,000 the classification provided in the respective articles:

Figure 1 Number of individual antibody tests reviewed in the present analysis

Focusing on the frequency of neuronal surface antibody (NSAb) in healthy controls, almost 26,500 individual antibody tests were performed. Of these, NSAb-positive samples are very rare (0.23%), indicating high disease specificity and supporting their pathogenic role and diagnostic usefulness in clinical routine. The majority of the reviewed 743,299 antibody tests for 22 NSAbs had to be excluded from analysis, as the respective articles did not report on the clinical phenotype of their participants, used assays without defined cutoffs or nonstandard assays, or did not specify whether the test was performed on either serum or CSF.

2 Neurology: Neuroimmunology & Neuroinflammation 1. Patients, defined by having symptoms for which NSAbs are RESULTS Using the search strategy, more than accepted to be specific, such as anti-NMDA receptor enceph- 1,000 articles on novel NSAbs were screened. From alitis; this group was sub-categorized into 2 cohorts, i.e., those these, 309 articles included NSAb testing in healthy with a highly suspicious clinical picture (fulfilling accepted and disease controls (table 1). The publication dates diagnostic criteria, if available) and those with a broader clin- ranged from 1974 for nAChR8 to October 2016. ical spectrum (table e-2). 2. Healthy participants, defined by exclusion of disease, usually In total, 30,485 serum and CSF samples of via questionnaires. In studies often referred to as “healthy,” healthy controls (HCs), 84,520 of disease controls, “normal,” or “blood donors.” and 18,441 samples of patients were tested for 1 of 3. Disease controls, containing multiple CNS disorders (table the 22 antibodies. A total of 609,853 tests were e-1), further subdivided into derived from patients for whom no clinical informa- a. Inflammatory other neurologic diseases (OND): other tion has been provided (figure 1). The numbers neurologic diseases with inflammation, such as demye- linating disorders and brain infections, include serum samples, CSF samples, and samples b. Noninflammatory OND: other noninflammatory neu- that are unspecified for serum or CSF by the authors. rologic diseases, such as degenerative disorders, neuro- From all HC samples, 99.9% were serum speci- logic tumors, and epilepsy, men. Most samples of HCs were tested for MOG c. Inflammatory other diseases (OD): other nonneurolog- antibodies (n 5 3,658), including n 5 28 CSF sam- ical diseases with inflammation, such as rheumatoid ples, least for neurexin-3a antibodies (n 5 21) and diseases and systemic infections, d. Noninflammatory OD: other nonneurological diseases none for igLON5 (for details, see table 1). without inflammation, such as malignancies, Regarding disease controls, 92% (n 5 77,775) of e. Psychiatric disorders. all tests included serum samples, and 4% were CSF (n

Table 1 Numbers of included articles and tested samples per antibody

HCs Disease controls Patients

Antigen Articles Serum CSF NS Serum CSF NS Serum CSF NS

AMPAR1/2 28 2,172 ——5,245 113 252 22 22 129

AMPAR3 9 399 ——1,149 ——123 ——

AQP4 82 3,065 ——8,864 489 48 4.988 135 61

D2R 13 232 ——868 13 14 81 ——

DNER 6 1,776 ——3,550 ——55 2 —

DPPX 12 1,753 ——3,465 12 464 118 3 —

GABAaR 12 285 ——1,660 203 589 276 160 57

GABAbR 23 1,836 ——4,768 153 141 91 15 —

GlyR 23 1,960 ——5,688 135 368 359 86 45

IgLON5 3 —————298 10 10 —

mGluR5 10 1,703 ——3,460 1 128 2 1 —

MOG 52 3,630 28 — 4,934 328 50 4.407 643 109

nAChR, m 25 877 ——3,687 10 — 819 ——

nAChR, g 15 1,073 ——2,227 10 — 306 ——

Neurexin-3a 121——— — 179 5 5 —

Neurofascin 12 433 ——905 ——2.114 2 —

NMDAR/NR1 64 3,204 7 — 8,249 1,564 539 518 336 334

NMDAR/NR2 11 124 — 905 86 133 259 53 —

PRG5 197——206 ——49 1 —

VGKC 44 1,366 ——5,034 226 115 372 29 180

Caspr2 37 2,238 ——6,640 28 14 405 — 147

Lgi1 36 2,206 ——6,271 28 14 344 — 153

Total 309 30,450 35 — 77,775 3,399 3,346 15,723 1,503 1,215

Abbreviations: HC 5 healthy control; NMDAR 5 NMDA receptor; NS 5 not specified. The numbers of reported neuronal surface antibody tests vary widely depending on the specific antibody, just as the numbers of publications. Of 30,485 available samples from HCs, only 35 were CSF specimens. When a publication contained tests for different antibodies in the same subject, they were counted here as individual tests.

Neurology: Neuroimmunology & Neuroinflammation 3 5 3,399) or remained unspecified for either serum or immunofluorescence-tagged secondary antibodies. Ear- CSF (n 5 3,346). The largest number of disease lier descriptions of antibodies showed a broader reper- controls was tested (n 5 10,352) for NMDAR/ toire of test methods, possibly related to a lack of NR1 antibodies, and the smallest number with n 5 standardization. By contrast, testing for more recent 179 samples for neurexin-3a (table 1). NSAbs, such as IgLON5, mGluR5, neurexin-3a, and Among tested samples categorized as “patients,” PRG5, was generally restricted to CBAs (table 2). 85% (n 5 15,723) were serum samples, only 8.2% In fact, the CBA has become the standard method (n 5 1,503) CSF, and 6.6% (n 5 1,215) were not for most NSAbs and is widely available in commercial specified. The number of tested patient samples laboratories.9,10 CBAs, comprising assays with fixed as ranged from n 5 3 for mGluR5 to n 5 5,184 for well as live cells, turned out to have superior sensitiv- AQP4 antibodies (table 1). ity for detection of NSAbs since they are specific for 1 antigen and its actual conformation is preserved.11 Various antibody assays. A wide variety of laboratory For MOG antibodies, for instance, the binding to methods have been used to detect NSAbs across studies, in particular ELISA, Western blots (WBs), radioimmu- nonconformational MOG, as in WB or ELISA, is noassays (RIAs), fluorescence immunoprecipitation as- not specific for demyelinating diseases and does not says, immunohistochemistry on sections of rodent correlate with the binding to conformational MOG brain, fluorescence-activated cell sorting (FACS), and in CBA or FACS on cells transfected with human cell-based assays (CBA) with visual evaluation of MOG.12 Furthermore, control groups may interfere

Table 2 Comparison of the diagnostic standard method and other methods for testing healthy control and disease control samples

Standard method Other methods

Antigen Controls (n) Positive (%) Controls (n) Positive (%)

AMPAR1/2 CBA 7,422 6 (0.1) IHC, WB 156 18 (11.5)

AMPAR3 239 0 ELISA, IHC, WB 964 118 (12.2)

AQP4 7,692 103 (1.3) ELISA, FACS, FIPA, IHC, RIA, WB 7,346 134 (1.8)

D2R 130 0 FACS, IHC, RIA 507 25 (4.9)

DNER 5,326 2 (0.0) IHC 207 0

DPPX 5,694 4 (0.1) IHC, WB 143 0

GABAaR 2,687 34 (1.3) IHC, WB 76 21 (27.6)

GABAbR 6,653 15 (0.2) ELISA, IHC 294 27 (9.2)

GlyR 8,151 109 (1.3) ND

IgLON5 298 1 (0.3) ND

mGluR5 5,292 2 (0.0) ND

MOG 5,209 20 (0.4) ELISA, FACS, RIA, WB 3,641 364 (10.0)

Neurexin-3a 200 0 ND

Neurofascin 398 4 (1.0) ELISA, FACS 1,080 15 (1.4)

NMDAR/NR1 13,136 140 (1.1) ELISA, FIPA, IHC, WB 1,197 45 (3.8)

NMDAR/NR2 104 0 ELISA, IHC, WB 1,144 199 (17.4)

PRG5 303 0 ND

Caspr2 8,799 122 (1.4) ELISA, FIPA, IHC 239 0

Lgi1 8,476 100 (1.2) ELISA, IHC 74 1 (1.4)

nAChR (m) RIA 4,288 73 (1.7) CBA, ELISA, Other 259 31 (12.0)

nAChR (g) 3,124 123 (3.9) LIPS 186 9 (4.8)

VGKC 6,227 318 (5.1) CBA, IHC 598 51 (8.5)

Total 99,848 1,176 (1.18) 18,111 1,058 (5.8)

Abbreviations: CBA 5 cell-based assay; FACS 5 fluorescence-activated cell sorting; FIPA 5 fluorescence immunoprecip- itation assays; IHC 5 immunohistochemistry; LIPS 5 luciferase-immunoprecipitation assay; ND 5 not done; NMDAR 5 NMDA receptor; RIA 5 radioimmunoassay; WB 5 Western blot. Different laboratory methods have been used to detect neuronal surface antibodies. Of these, CBA (and in few tests RIA) has become the standard method because of superior sensitivity and commercial acceptance and dissemination.

4 Neurology: Neuroimmunology & Neuroinflammation with assay interpretation, such as in hypergammaglo- with a positivity rate of 6.1% in 4,220 samples. Sam- bulinemia or rheumatoid diseases13; they are therefore ples from disease controls tested for IgLON5 and provided separately in detail (table e-1). neurexin-3a were not specified for either serum or Therefore, throughout this review, we define the CSF and are therefore not listed in table e-3. One of CBA as “standard method” for all NSAbs except for 298 noninflammatory OND was positive for antibod- VGKC and nAChR antibodies, for which the stan- ies against IgLON5, but none of 179 disease controls dard method is the RIA (table 2). For AQP4, ELISA for neurexin-3a antibodies (table e-1). and FACS are still frequently used,13 but considered In contrast to healthy and disease controls, pa- nonstandard and thus grouped together with higher tients in which an NSAb-mediated disease was sus- seropositivity (2.4%) compared with CBA (1.3%). pected on clinical grounds indeed had profoundly Although the detection of VGKC antibodies with higher prevalences. NSAbs were detected in 21.0% RIA has currently been replaced by the CBA for the of 4,832 samples in the group with less specific clin- disease-specific target antigens Lgi1 and Caspr2,14 the ical symptoms, while patients with a highly suggestive still wide availability of testing prompted us to clinical picture were NSAb-positive in 69.2% of include VGKC complex antibodies (with and with- 3,185 serum samples. out Lgi1/Caspr2 antibodies) to provide clinicians Limited data on antibody frequencies in CSF. Given the with antibody frequencies (table e-1). invasive nature of a lumbar puncture, almost no data Overall, 1.2% of blood and CSF samples from are available on CSF antibodies in HCs. One study controls (i.e., suspected of not having an antibody- reported 7 samples acquired in the diagnostic workup mediated disease) were positive with the standard of headache; all were negative for NMDAR/NR1 method. By contrast, 5.8% of the control samples antibodies (table e-4).15 As also examined for head- yielded a positive result if tested with nonstandard ache, 1 of 28 HCs was positive in CSF for MOG methods, suggesting a markedly higher rate of false- antibodies tested by ELISA.16 positives (table 2). From disease controls 2,781 CSF samples have All following analyses of the NSAbs prevalence been tested with the standard method for 14 of 22 will consider samples tested with the standard meth- antibodies. Forty-four (1.6%) positive samples were ods only. Samples tested with other methods (e.g., identified for AQP4, GlyR, NMDAR/NR1, and ELISA not providing a clear cutoff for positivity; VGKC antibodies. From these, 35 belonged to the figure 1) are categorized in table e-1. inflammatory OND cohort; 8.2% (of 135 samples) were positive for GlyR, 5.3% (of 19) for VGKC, NSAbs are very rare in healthy participants. Serum anal- 2.1% (of 468) for AQP4, and 1% (of 1,551) positive yses revealed a low frequency of NSAbs in HCs, on for NMDAR/NR1 antibodies. No CSF was available average 0.23% (n 5 61) of 26,423 antibody tests for AMPAR3, DNER, neurofascin, NMDAR/NR2, (table e-3). In detail, 0% positivity was seen for 9 and PRG5 (table e-4). different antibodies including 8,822 tests, namely In contrast to healthy and disease controls, pa- for AMPAR3, AQP4, DNER, GABAaR, GABAbR, tients with suspected NSAb-mediated symptoms mGluR5, nAChR (m), neurexin-3a, and PRG5. On had indeed a higher prevalence of CSF antibodies. average, 1,201 samples were tested per antibody. No However, the prevalence of NSAbs largely depended tests on HCs were reported yet with CBA for D2R, on the cohort size and selection. The highest preva- IgLON5, and NMDAR/NR2 (tables e-1 and e-3). lence of 100% is reached in the smallest cohorts with For comparison, the frequency in disease controls an average of only 4.7 CSF samples per antibody; this was somewhat higher with 1.5% (n 5 989) of 67,069 includes antibodies against DNER, DPPX, tests, depending on the clinical subgroup. NSAbs in GABAbR, neurexin-3a, neurofascin, and PRG5. the psychiatric cohort (0.25% of 28,491) were simi- Altogether, patients with a highly suggestive clinical larly common as in HCs. NSAbs occurred with picture were CSF NSAb-positive in 76.9% (of 555 increasing frequency in inflammatory OD (0.34% samples) and patients with a less distinctive clinical of 888 tests), noninflammatory OND (1.76% of picture still in 26.3% (of 323 samples) (table e-4). 17,718), noninflammatory OD (2.18% of 6,889), and inflammatory OND (3.46% of 13,083) (table Most studies contained small numbers of controls. It e-3). Noteworthy is the high prevalence in nonin- became obvious from the literature research on flammatory OD, which likely relates to underlying NSAbs that most studies tested only small numbers malignancies in 76.7% of NSAb-positive noninflam- of control samples (figure e-1). In fact, only 1 study matory OD patients. included a large number of HCs and might thus No positive antibodies in disease controls were seen dominate the interpretation of antibody frequencies; for AMPAR3, D2R, NMDAR/NR2, and PRG5. By HCs were blood donors presuming that they are contrast, VGKC antibodies were frequently detected the best surrogate that is available.17 In this study,

Neurology: Neuroimmunology & Neuroinflammation 5 1.703 HC samples were analyzed for antibodies given that it clearly distinguishes patients with typical against AMPAR1/2, AQP4, DNER, DPPX, symptoms of NSAb-associated diseases in which GABAbR, GlyR, mGluR5, MOG, NMDAR/NR1, 69.2% were antibody-positive in serum, certainly with Caspr2, and Lgi1, however, restricted to serum. Zero broad differences among groups. Thus, NSAbs are well positive samples were found for AQP4, DNER, qualified as laboratory markers for antibody-mediated GABAbR, and mGluR5. The most frequent antibody neuropsychiatric diseases. To determine sensitivity and was targeting NMDAR/NR1 in 20 cases (1.2%). The specificity on clinical grounds, comparison with closely second largest HC cohort contained 274 samples, all related disease groups is nonetheless required. negative for NMDAR/NR1.18 Cohorts exceeding 100 HC samples were only available for 8 of 22 anti- Prospective larger studies are needed and should include bodies (figure e-1). Frequency determination of CSF samples. Analysis of the data brought to light that mGluR5 antibodies in HCs was restricted to a single the knowledge on antibody prevalence in HCs is still study using CBA.17 scarce compared with the dramatic growth in NSAb- The marked lack of systematic studies on antibody mediated diseases. The main limitations in the field prevalence in HCs was less striking for disease con- for interpreting the data are (1) the often very small trols. Three studies each described more than 1,000 sample number of controls in studies publishing samples, including psychiatric disorders and nonin- a novel antibody and (2) the lack of CSF controls flammatory OND (n 5 2,533; 0%–1.34% positive given that the CSF antibodies are often more disease- for 10 NSAbs),17 people older than 60 years with relevant than serum. unspecified diseases tested for muscle nAChR (n 5 The lack of CSF samples from controls makes the 2,000; 0.5% positive),19 and MS patients tested for determination of specificity difficult in cases where AQP4 antibodies (n 5 1,040; 0.3% positive, diagno- CSF antibodies predominate, such as for NMDAR/ sis in 2 of 3 was revised to NMOSD).20 In a recent NR1, while it is less important when serum testing study of 925 acutely ill psychiatric patients, 1.4% is generally sufficient, such as for MOG and AQP4. were NSAb-positive: 1 for AMPAR1/2, 5 for In fact, almost 95% of all reported control samples NMDAR/NR1, and 7 for Caspr2 antibodies.21 A were serum probes. CSF samples were tested with subgroup of these patients was analyzed for DNER, the standard method from 2,788 controls, but for 8 DPPX, GlyR, mGluR5, and MOG antibodies, which antibodies, not a single control CSF has been tested. were negative.22 All further studies contained mark- Despite a relatively high number of studies that 5 edly smaller cohorts with n 5 98 for PRG523 to n 5 tested controls for the presence of NSAb (n 309), 741 tests for GlyR antibodies24 (mean 401 antibody only very few reported the systematic collection of tests per largest cohort), the positive cases ranged samples, and most studies dealt with no or low num- 17,19–21 from 0% to 4.1% (mean 0.9%) (table e-1). Meaning- bers of HCs. Data from the few exceptions indi- ful control groups were not always provided, such as cated that NSAbs are very infrequent in healthy and patients with isolated lung cancer to compare with disease controls, ranging from 0% to 1.3% (mean PRG5 antibody-positive cerebellar degeneration or 0.23%). It is therefore possible that such small percent- patients with Hodgkin’s lymphoma to compare with age of antibody-positive participants can easily be over- mGluR5 antibody-positive encephalitis. looked in studies with control numbers below 100. Another possibility for the low rate of NSAb-positive DISCUSSION Focusing on the prevalence of the 22 cases in most studies might be susceptibility to selec- most common antineuronal autoantibodies in tion bias or the lack of blinding to controls in the healthy and disease controls, we identified 115,005 experimental design. On the contrary, it is possible antibody tests performed between 1974 and 2016. that the true prevalence in HCs is even lower, given Of these, 96,280 samples were tested with the current that most samples are from blood donors in which an standard method, which is the CBA for most NSAbs extensive screen for comorbidities is not feasible. except RIA for VGKC and nAChR antibodies. Based on the comprehensive analysis, the frequency of Interpretation of a positive test result—6 things to consider NSAbs in HCs was very low with only 0.23% in atypical clinical constellations. In clinical practice, antibody-positive participants. Some NSAbs were the search for NSAb has become routine and broad not detected in a single HC subject, including laboratory panels occasionally return positive AMPAR3, AQP4, DNER, GABAaR, GABAbR, NSAbs in patients for which they were not initially mGluR5, nAChR (m), neurexin-3a, and PRG5 an- suspected. To interpret the positive test result cor- tibodies. For comparison with a well-known non- rectly, one has to take the following aspects into NSAb, antibodies against thyroid peroxidase were consideration. positive in 11.3% of 16,533 controls.25 The very low First, although rare with the state-of-the-art CBAs, prevalence of NSAbs in HCs suggests high specificity, the test might represent a false-positive result.26 The

6 Neurology: Neuroimmunology & Neuroinflammation validity can be markedly improved if 2 different assays of most NSAbs is very low (0.23%) in HCs and well- are combined, e.g., the CBA plus determination of selected disease controls. Increasing experimental the characteristic anatomical binding to brain sections evidence for antibody pathogenicity will facilitate the using immunofluorescence, which should be aimed clinical decision for immunotherapy in NSAb- to prevent unnecessary immunotherapy. positive patients. Second, the NSAb might be present in the circu- lation (true-positive), but not reaching its target in the AUTHOR CONTRIBUTIONS brain or the peripheral nervous system. In that case, Katharina Lang: drafting/revising the manuscript; analysis or interpreta- other factors might be relevant for pathogenic anti- tion of data; and acquisition of data. Harald Prüss: drafting/revising the body effects (“second hit”), including the integrity manuscript; study concept or design; analysis or interpretation of data; acquisition of data; statistical analysis; study supervision; and obtaining of the blood-brain barrier or intrathecal antibody funding. production. Third, NSAb might hypothetically cause changes STUDY FUNDING that can be measured by MRI or electrophysiology No targeted funding reported. but are not obvious on clinical examination. Future studies will have to determine clinical relevance for DISCLOSURE neurologic patients, similar to other antibody- The authors report no disclosures. Go to Neurology.org/nn for full dis- mediated disorders such as celiac disease, where severe closure forms. duodenal lesions frequently occur in asymptomatic patients.27 Received April 1, 2017. Accepted in final form May 25, 2017. Fourth, the antibody could bind its target (and REFERENCES thus be positive in the diagnostic assay) but have no 1. Dalmau J, Tuzun E, Wu HY, et al. Paraneoplastic anti- pathogenic effect. N-methyl-D-aspartate receptor encephalitis associated Fifth, clinical relevance also relates to the knowl- with ovarian teratoma. Ann Neurol 2007;61:25–36. edge on the pathogenicity of NSAbs. For many 2. Hughes EG, Peng X, Gleichman AJ, et al. Cellular and NSAbs, neuronal damage has been demonstrated synaptic mechanisms of anti-NMDA receptor encephalitis. – in vitro and in vivo,2,5,28 including human monoclo- J Neurosci 2010;30:5866 5875. 3. Kreye J, Wenke NK, Chayka M, et al. Human cerebro- nal autoantibodies from CSF-derived plasma cells.3 In spinal fluid monoclonal N-methyl-D-aspartate receptor such cases, positive titers are likely risk factors for autoantibodies are sufficient for encephalitis pathogenesis. neuropsychiatric symptoms, similar to a drug or toxin Brain 2016;139:2641–2652. level.3 4. Planaguma J, Haselmann H, Mannara F, et al. Ephrin-B2 Sixth, a positive NSAb in a patient not initially prevents N-Methyl-D-Aspartate receptor antibody effects suspected to have an antibody-mediated disease can on memory and neuroplasticity. Ann Neurol 2016;80: 388–400. still be “true-positive” and represent the further ex- 5. Planaguma J, Leypoldt F, Mannara F, et al. Human panding phenotypic spectrum seen in most N-methyl D-aspartate receptor antibodies alter memory subgroups. and behaviour in mice. Brain 2015;138:94–109. Having these possibilities in mind, bedside inter- 6. Lai M, Hughes EG, Peng X, et al. AMPA receptor anti- pretation of a positive antibody result must integrate bodies in limbic encephalitis alter synaptic receptor loca- – antibody prevalence data, careful clinical evaluation tion. Ann Neurol 2009;65:424 434. of alternative diagnoses, predominance in CSF vs 7. Graus F, Titulaer MJ, Balu R, et al. A clinical approach to diagnosis of autoimmune encephalitis. Lancet Neurol serum, experience of the laboratory performing the 2016;15:391–404. diagnostic assay, and consideration of antibody tests 8. Almon RR, Andrew CG, Appel SH. Serum globulin in by a different method. myasthenia gravis: inhibition of alpha-bungarotoxin bind- ing to acetylcholine receptors. Science 1974;4:55–57. CONCLUSIONS NSAbs are rarely found in HCs, 9. Stocker W, Saschenbrecker S, Rentzsch K, Komorowski L, suggesting high disease specificity and supporting Probst C. Autoantibody diagnostics in neurology using native and recombinant antigenic substrates [in German]. their pathogenic role and diagnostic usefulness in Nervenarzt 2013;84:471–476. clinical routine. Conversely, positive titers in atypical 10. Lancaster E. The diagnosis and treatment of autoimmune patients might point to the still expanding phenotypic encephalitis. J Clin Neurol 2016;12:1–13. spectrum, in particular if confirmed in 2 independent 11. van Coevorden-Hameete MH, Titulaer M, Schreurs diagnostic assays. Future prospective studies are MWJ, Hoogenraad C. Detection and characterization of needed and should include more CSF samples and autoantibodies to neuronal cell-surface antigens in the cen- tral nervous system. Front Mol Neurosci 2016;9:37. data from HCs measured with blinded CBAs and dif- 12. Lalive PH, Menge T, Delarasse C, et al. Antibodies to ferent assay systems. Control groups in articles pub- native myelin oligodendrocyte glycoprotein are serologic lishing novel autoantibodies should include at least markers of early inflammation in multiple sclerosis. Proc 300–400 participants, given that the mean prevalence Natl Acad Sci USA 2006;103:2280–2285.

Neurology: Neuroimmunology & Neuroinflammation 7 13. Fryer JP, Lennon VA, Pittock SJ, Jenkins SM, Wingerchuk 21. Schou M, Saether SG, Borowski K, et al. Prevalence of DM, McKeon A. AQP4 autoantibody assay performance in serum anti-neuronal autoantibodies in patients admitted to clinical laboratory service. Neurol Neuroimmunol Neuroin- acute psychiatric care. Psychol Med 2016;46:3303–3313. flamm 2014;1:e11. doi: 10.1212/NXI.0000000000000011. 22. Sæther SG, Schou M, Stoecker W, et al. Onconeural anti- 14. van Sonderen A, Petit-Pedrol M, Dalmau J, Titulaer MJ. bodies in acute psychiatric inpatient care. J Neuropsychiatry The value of LGI1, Caspr2 and voltage-gated potassium Clin Neurosci 2017;29:74–76. channel antibodies in encephalitis. Nat Rev Neurol 2017; 23. van Coevorden-Hameete MH, de Graaff E, Titulaer MJ, 13:290–301. et al. Plasticity-related gene 5: a novel surface autoantigen 15. Doss S, Wandinger KP, Hyman BT, et al. High preva- in paraneoplastic cerebellar degeneration. Neurol Neuro- lence of NMDA receptor IgA/IgM antibodies in different immunol Neuroinflamm 2015;2:e156. doi: 10.1212/NXI. dementia types. Ann Clin Translational Neurol 2014;1: 0000000000000156. 822–832. 24. Carvajal-Gonzalez A, Leite MI, Waters P, et al. Glycine 16. Karni A, Bakimer-Kleiner R, Abramsky O, Ben-Nun A. receptor antibodies in PERM and related syndromes: char- Elevated levels of antibody to myelin oligodendrocyte gly- acteristics, clinical features and outcomes. Brain 2014;137: coprotein is not specific for patients with multiple sclerosis. 2178–2192. Arch Neurol 1999;56:311–315. 25. Hollowell JG, Staehling NW, Flanders WD, et al. Serum 17. Dahm L, Ott C, Steiner J, et al. Seroprevalence of auto- TSH, T4, and thyroid antibodies in the United States antibodies against brain antigens in health and disease. population (1988 to 1994): National Health and Nutri- Ann Neurol 2014;76:82–94. tion Examination Survey (NHANES III). J Clin Endocri- 18. Busse S, Busse M, Brix B, et al. Seroprevalence of N-methyl- nol Metab 2002;87:489–499. D-aspartate glutamate receptor (NMDA-R) autoantibodies 26. Gastaldi M, Thouin A, Franciotta D, Vincent A. in aging subjects without neuropsychiatric disorders and in Pitfalls in the detection of N-methyl-d-aspartate- dementia patients. Eur Arch Psychiatry Clin Neurosci 2014; receptor (NMDA-R) antibodies. Clin Biochem 2017; 264:545–550. 50:354–355. 19. Vincent A, Clover L, Buckley C, Rothwell PM. Evidence 27. Trovato CM, Montuori M, Anania C, et al. Are ES- of underdiagnosis of myasthenia gravis in older people. PGHAN “biopsy-sparing” guidelines for celiac disease also J Neurol Neurosurg Psychiatry 2003;74:1105–1108. suitable for asymptomatic patients? Am J Gastroenterol 20. Pittock SJ, Lennon VA, Bakshi N, McKeon A, Barcellos 2015;110:1485–1489. LF. Seroprevalence of aquaporin-4–IgG in a Northern 28. Pettingill P, Kramer HB, Coebergh JA, Pettingill R, California population representative cohort of multiple Waters P, Vincent A. Antibodies to GABAA receptor a1 sclerosis. JAMA Neurol 2014;71:1433–1436. and g2 subunits. Neurology 2015;84:1233–1241.

8 Neurology: Neuroimmunology & Neuroinflammation Clinical/Scientific Notes

Michael Bonello, MRCP IgLON5 DISEASE RESPONSIVE TO to a rapid improvement over a few weeks. A repeat Anu Jacob, FRCP, DM IMMUNOTHERAPY CSF analysis confirmed raised protein suggestive of Mark A. Ellul, MRCP persistent inflammation, and further IV-pulsed cyclo- Erandi Barker, MRCP phosphamide was given. After 2 pulses, his behavior A 45-year-old man was seen with a history of confu- Robert Parker, FRCP, normalized and sleep pattern improved with return sion and disorientation for 1 year, during which time FFICM of dreams, and resolution of neck pain and dysphagia. he was unable to identify relatives. He was unable to Samantha Jefferson, Serum antibodies to IgLON5 returned positive cope at his work as a plasterer, noticed neck pain, dys- MRCP, PhD using indirect immunofluorescence (Euroimmun). phagia, and unexplained weight gain. His family re- Sundus Alusi, FRCP, MD CSF testing for the antibody was also positive ported that his sleep pattern had changed, describing (figure e-3). The patient continued treatment with his sleep as disturbed with episodes consistent with Neurol Neuroimmunol cyclophosphamide, and he has received 8 pulses Neuroinflamm stridor, myoclonus (the video at Neurology.org/nn), so far. His human leukocyte antigen (HLA) genotyp- 2017;4:e383; doi: 10.1212/ and semipurposeful movements. NXI.0000000000000383 ing confirmed HLA-DQB1*05:01 and HLA- At his initial examination, he appeared mildly con- DRB1*10:01 alleles. fused with a depressed affect. He was obese (body At his last review (2 years from the onset of his mass index of 33 kg/m2) and was noted to be centrally initial symptoms), he continues to improve. He cyanosed (figure e-1). He had mild gait ataxia, bulbar exhibited no evidence of cognitive impairment or dysarthria, and dysphagia. The rest of his examination abnormal behavior, and there were no involuntary was normal. He was found to have established venti- movements. His gait remains mildly ataxic. Noninva- latory failure and sleep-disordered breathing. EEG sive ventilation has been established with improve- demonstrated a baseline of theta waves with a rhythm ment in his arterial blood gases. His spouse reports intermixed. CSF analysis was abnormal, suggesting better sleep patterns. inflammation (table). MRI of the head and neck was normal. Inpatient polysomnography with video was extremely limited by the patient’s limited sleep Discussion. A recent report suggested treatment duration and nocturnal behavioral problems. The with immunotherapy for a patient with IgLON5 awake period evaluation revealed intermittent runs encephalitis.1 Here, we provide further evidence of of theta wave activity suggesting an increased homeo- an IgLON5-associated disorder that has shown sus- static sleep drive. Frequent spontaneous desaturations tained response to immunotherapy. were noted, some of which were central in nature. IgLON5 antibody–associated encephalopathy was The sleep period evaluation demonstrated an first described in 2014.2 Sleep disturbance was increased sleep onset latency and reduction in non- characteristic in all these patients’ presentations. REM stage 2 sleep and a complete absence of REM Other features including gait ataxia, bulbar dysar- sleep (figure e-2). His apnea-hypopnea index was 25, thria, and dysphagia2 were also present. Movement confirming moderate obstructive sleep apnea. disorders associated with IgLON5 syndrome, Nocturnal noninvasive ventilation was issued for including orofacial and limb chorea,1 dystonia, hy- long-term use, but initial adherence was poor. pomimia, bradykinesia, and myoclonus.3 Cognitive The patient was initially treated with prednisolone decline featuring impaired executive function, vi- and immunoglobulins. This was associated with suospatial dysfunction, and episodic memory loss has improvement in behavior mirrored by improvement been reported.4 Brain imaging is typically normal. Supplemental data at in CSF parameters, and he became well enough to The youngest patient reported in the literature was 52 Neurology.org/nn be discharged home. However, he was admitted 6 years old.2 weeks later with aspiration pneumonia and ventila- The physiologic role of IgLON5 is unknown, but tory failure requiring intensive care support. His sleep other members of the IgLON family are involved in disorder was still prominent clinically and required synaptic and neuronal formation during brain devel- further treatment with plasmapheresis followed by opment.5 Antibodies to IgLON5 have been originally another course of IV immunoglobulins, which led linked with a tauopathy when they were detected in 8

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Acknowledgment: The authors thank Edward Jackson and the staff Table CSF analysis confirmed a pleocytosis (100% lymphocytes) with raised in the Neurosciences Laboratories with the help processing and pro- protein suggestive of CSF inflammation viding the immunofluorescence results. Study funding: No targeted funding reported. December 2015 January 2016 March 2016 July 2016 Disclosure: M. Bonello reports no disclosures. A. Jacob serves on the Red blood cell count, per mm3 0 1,920 3 4 scientific advisory board for Shire Pharmaceuticals; received speaker honoraria from Biogen Idec, Shire, and Terumo BCT; consulted for White blood cell count, per mm3 14 3 1 1 Shire Pharmaceuticals; and received research support from Biogen, Total protein, g/L 0.84 0.77 0.82 0.84 Alexion, Shire, NHS, and University of Liverpool. M.A. Ellul

CSF glucose, mmol/L 4.0 6.5 6.6 6.2 received research support from the Association of British Neurologists. E. Barker, R. Parker, and S. Jefferson report no disclosures. S. Alusi Plasma glucose, mmol/L 3.5 5.8 6.9 6.4 received travel funding from Boston Scientific. Go to Neurology.org/nn Oligoclonal bands Negative Negative Negative Negative for full disclosure forms. The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives Li- cense 4.0 (CC BY-NC-ND), which permits downloading and shar- 2 patients with a similar clinical presentation. ing the work provided it is properly cited. The work cannot be IgLON5-associated disorder provides an interesting changed in any way or used commercially without permission from link between neurodegeneration and autoimmunity. the journal. All patients who were HLA genotyped had the same Received March 13, 2017. Accepted in final form June 2, 2017. alleles: HLA-DQB1*05:01 and HLA-DRB1*10:01.2 Correspondence to Dr. Bonello: Michael.bonello@thewaltoncentre. Autopsy on 6 patients revealed hyperphosphorylated nhs.uk tau protein deposited in the hypothalamus, prehypo- thalamic region, the tegmentum, and the upper 1. Haitao R, Yingmai Y, Yan H, et al. Chorea and parkin- cervical cord.6 The presence of a pleocytosis on CSF sonism associated with autoantibodies to IgLON5 and and improvement following immunotherapy in our responsive to immunotherapy. J Neuroimmunol 2016; 300:9–10. patient points toward a complex interplay of autoim- 2. Sabater L, Gaig C, Gelpi E, et al. A novel non-rapid-eye munity, genetic predisposition, and neurodegenera- movement and rapid-eye-movement parasomnia with sleep tion. Furthermore, recent evidence exploring the breathing disorder associated with antibodies to IgLON5: mechanisms of action of IgLON5 in rat hippocampal a case series, characterisation of the antigen, and post- neurons suggests that antibodies decreased cell surface mortem study. Lancet Neurol 2014;13:575–586. IgLON5 clusters with internalization of antibody not 3. Hogl B, Heidbreder A, Santamaria J, Graus F, Poewe W. IgLON5 autoimmunity and abnormal behaviours during reversed once the IgLON5 antibodies were removed sleep. Lancet 2015;385:1590. 7 from the media. This suggests a pathogenic role of 4. Simabukuro MM, Sabater L, Adoni T, et al. Sleep disorder, these antibodies in the disease and raises the possibil- chorea, and dementia associated with IgLON5 antibodies. ity of a treatable phase of the disease possibly in the Neurol Neuroimmunol Neuroinflamm 2015;2:e136. early stages. 5. Sanz R, Ferraro GB, Fournier AE. IgLON cell adhesion molecules are shed from the cell surface of cortical neurons From the Department of Neurology (M.B., A.J., M.A.E., S.A.), and to promote neuronal growth. J Biol Chem 2015;290: Department of Neurophysiology (S.J.), The Walton Centre NHS 4330–4342. Foundation Trust; and Department of Respiratory and Sleep Medi- cine (E.B., R.P.), Aintree University Hospital NHS Foundation 6. Gelpi E, Hoftberger R, Graus F, et al. Neuropathological Trust, Liverpool, United Kingdom. criteria of anti-IgLON5-related tauopathy. Acta Neuropa- – Author contributions: M.B.—primary author and looked after the thol 2016;132:531 543. patient. A.J., M.A.E., E.B., and R.P.—contributed to the manuscript 7. Sabater L, Planaguma J, Dalmau J, Graus F. Cellular and involved in patient’streatment.S.J.—contributed to the manu- investigations with human antibodies associated with script and neurophysiologist facilitating sleep studies. S.A.—lead the anti-IgLON5 syndrome. J Neuroinflammation consultant looking after the patient and contributor to the manuscript. 2016;13:226.

2 Neurology: Neuroimmunology & Neuroinflammation Clinical/Scientific Notes

Giovanni Novi, MD DRAMATIC REBOUNDS OF MS DURING clinically (EDSS 7.0) and radiologically (figure, C). Angelo Ghezzi, MD PREGNANCY FOLLOWING FINGOLIMOD One month later, her clinical condition deteriorated Matteo Pizzorno, MD WITHDRAWAL (EDSS 8.0) with worsening of brain MRI (figure, D). Caterina Lapucci, MD She was treated again with IV cyclophosphamide 2 Fabio Bandini, MD In MS, discontinuation of some therapies may result (1 g/m ), but a few days later, she developed septic Pietro Annovazzi, MD in a transitory, possibly overwhelming, increase in shock and died. Giovanni L. Mancardi, disease activity.1–4 We report 2 cases of severe clinical The second patient was diagnosed with MS in MD, PhD rebound after fingolimod suspension occurring in 2 1995 and was treated from 1999, first with inter- Antonio Uccelli, MD, patients following pregnancy. Clinical rebound was feron-b1a once weekly, then interferon-b1a 22 mcg PhD defined as the occurrence of new severe neurologic 3 times weekly and subsequently interferon-b1a 44 symptoms together with a significant increase of new mcg 3 times weekly. In 2004, because of insufficient Neurol Neuroimmunol disease control, she received 4 infusions with mitox- Neuroinflamm or enlarging T2-weighted or gadolinium-enhancing 2 2017;4:e377; doi: 10.1212/ T1-weighted lesions exceeding baseline activity, on antrone 12 mg/m monthly followed by interferon- NXI.0000000000000377 treatment discontinuation.4 The first patient was b1b. In May 2005, interferon-b1b was suspended diagnosed with MS in 2005. In the subsequent 3 when the patient resulted pregnant. In 2006, follow- years, she experienced several relapses and progressive ing childbirth, she started azathioprine and after a few increase of MRI lesion load despite the sequential months, because of disease progression, she was introduction of weekly interferon-b1a, thrice weekly infused with 4 additional doses of mitoxantrone. In interferon-b1a and then glatiramer acetate. Natalizu- 2007, she started NTZ with no signs of disease activ- mab (NTZ) was then started (EDSS 3.0, JCV anti- ity for 28 months. In October 2010, NTZ was sus- bodies negative status), resulting in no evidence of pended fearing PML. In June 2011, she started disease activity for the following 40 months. On sero- fingolimod with incomplete disease control. In Sep- conversion, NTZ was interrupted in August 2011 tember 2014 (EDSS 3.0), she was again pregnant and beacause of the potential risk of progressive multifocal therapy was suspended. Lymphocyte count was 0.7 3 leukoencephalopathy (PML) and fingolimod was 109/L. Four weeks later, with a lymphocyte count of started. In March 2013, fingolimod was suspended 2 3 109/L, she experienced severe left leg monopa- as disease was stable (figure, A) and due to desire of resis and dysarthria (EDSS 6.0). IV methylpredniso- pregnancy and potential risks of fetal exposure to lone was administered with no results. Four weeks fingolimod.5 Lymphocyte count was 0.34 3 109/L. later, therapeutic abortion was performed. One week In April 2013, she resulted pregnant. In July (elev- later, she presented with tetraplegia and anarthria enth week of pregnancy), she abruptly developed (EDSS 9.5). She received high-dose steroids, followed quadriplegia and signs of severe cognitive impairment by plasma-exchange and rituximab (two 1-g infusions (EDSS: 9.0). Lymphocyte count was 1.54 3 109/L. 15 days apart) with radiologic and clinical improve- One gram of methylprednisolone was administered ment (EDSS 5.5 on April 2015). In May 2015, IV for 5 days with partial recovery (EDSS 7.0). dimethyl fumarate was started with slow progression One month later, she developed locked-in syndrome to EDSS 6.5 (January 2017). (EDSS 9.5). Brain MRI showed a dramatic increase Suspension of therapies affecting lymphocytes of white-matter abnormalities with many trafficking to the CNS, such as NTZ and fingolimod, gadolinium-enhancing lesions (figure, B). PML was may rarely result in disease rebound. Compared with excluded by CSF analysis. IV methylprednisolone dramatic deterioration of disease following NTZ sus- followed by IV immunoglobulins were administered pension, fingolimod withdrawal usually results in with no success. Plasma exchange was not considered mild disease recrudescence,3,4 mostly controlled by because of the possible adverse effects on maternal steroids. In the reported cases, rebound after fingoli- hemodynamics and limited experience on pregnant mod suspension was rapidly devastating despite patients with MS. On therapeutic abortion, she occurring during the early phase of pregnancy, a con- received IV cyclophosphamide (2 g/m2), improving dition associated with a significant reduction of

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure Patient 1 MRI scans

(A) Brain MRI performed on October 12, 2012. (A.a) FLAIR (FLuid Attenuated Inversion Recovery) sequence. (A.b) T1 gadolinium enhanced sequence. Routine MRI scan during fingolimod therapy shows some periventricular juxtacortical FLAIR white-matter hyperintensities with no gadolinium enhanced lesions. (B) Brain MRI performed on August 29, 2013. In the FLAIR sequence diffuse and confluent white-matter hyperintensities are found in both hemispheres in the periventricular and subcortical white matter, with the involvement of the “U” fibers. Many lesions show gadolinium enhancement. (C) Brain MRI performed on October 17, 2013. One month after cyclophosphamide IV administration. FLAIR white-matter hyperintensities are significantly reduced (C.a) and no lesion display gadolinium enhancement (C.b) compared with the MRI performed on 29th of August (B). (D) Brain MRI performed on January 3, 2014. A remarkable increase in the number and size of white-matter hyperintensities can be observed (FLAIR sequence, D.a) compared with the scan of 17th of October. New diffuse and rim gadolinium enhancing lesions are present (D.b).

relapse risk. Possibly, abortion may have led to hor- of possible risks of rapid disease worsening after monal changes contributing to increasing disease treatment suspension. activity similarly to what was observed in the post- From the Department of Neurology, Rehabilitation, Ophthalmology, partum. Genetics, Maternal and Child Health (DINOGMI) (N.G., C.L., Thus, we suggest that similar to current protocols G.L.M., A.U.), University of Genoa, Italy; Multiple Sclerosis Study Center (A.G., P.A.), Hospital of Gallarate, Italy; San Paolo Hospital adopted for NTZ discontinuation, fingolimod sus- (M.P., F.B.), Savona, Italy; and IRCCS AUO San Martino—IST pension in patients with high-disease activity during (G.L.M., A.U.), Genoa, Italy. previous treatments should be carefully monitored Author contributions: G. Novi: acquisition of data and study concept to detect early signs of disease reactivation poten- and design. A. Ghezzi: study supervision and critical revision of the manuscript for important intellectual content. M. Pizzorno and tially requiring immediate highly effective treat- C. Lapucci: acquisition of data. F. Bandini: study supervision and ments. Moreover, early lymphocytes levels increase critical revision of the manuscript for important intellectual content. after fingolimod cessation could represent a warning P. Annovazzi: acquisition of data. G.L. Mancardi and A. Uccelli: sign possibly associated with their migration to the study supervision and critical revision of the manuscript for impor- tant intellectual content. CNS.6 Whether these rare events are due to a possi- Study funding: No targeted funding reported. ble change in the expression of key S1P receptors on Disclosure: G. Novi reports no disclosures. A. Ghezzi served on the immune or CNS target cells will require further editorial board for Multiple Sclerosis International and served as studies.7 Thepossibleoutcomefollowingtherapy a speaker for Merck Serono, Novartis, Genzyme, Biogen, Teva; con- discontinuation should be carefully considered at sulted for Novartis, Biogen, Mylan, Roche. M. Pizzorno, C. Lapucci, and F. Bandini report no disclosures; P. Annovazzi served on advi- the time of treatment choice in young women in sory boards for Novartis, Genzyme, Teva, Roche, Merck-Serono, child-bearing age, who need to be properly advised Biogen and received speaker honoraria from Teva Italia, Novartis,

2 Neurology: Neuroimmunology & Neuroinflammation Genzyme, Merck-Serono, Biogen. G.L. Mancardi received travel following cessation of natalizumab therapy. Arch Neurol funding and/or speaker honoraria from Bayer Schering, Biogen, 2011;68:186–191. Merck Serono, Novartis, Sanofi-Aventis, Teva; is an associate editor 2. Sorensen PS, Koch-Henriksen N, Petersen T, Ravnborg M, for Neurological Sciences; and received research support from Italian Oturai A, Sellebjerg F. Recurrence or rebound of clinical Multiple Sclerosis Society. A. Uccelli served on the scientific advisory relapses after discontinuation of natalizumab therapy in board for Genetech, Roche, Allargan; received travel funding and/or highly active MS patients. J Neurol 2014;261:1170–1177. speaker honoraria from Athersys, Biogen Dompe, Serono Symposia, Sanofi-Aventis, Tohoku University Graduate School of Medicine; 3. Sempere AP, Berenguer-Ruiz L, Feliu-Rey E. Rebound of consulted for Merck-Serono, Biogen Dompe; and received research disease activity during pregnancy after withdrawal of fingo- support from Bayer Schering, Sanofi-Aventis, Italian Ministry of limod. Eur J Neurol 2013;20:e109–e110. Health, Regione Liguria Progetto Limonte, Ricera Sanitaria Region- 4. Hatcher SE, Waubant E, Nourbakhsh B, Crabtree-Hartman E, ale, Italiana Sclerosi Multipla, Fondazione CARIGE. Go to Graves JS. Rebound syndrome in patients with multiple scle- Neurology.org/nn for full disclosure forms. The Article Processing rosis after cessation of fingolimod treatment. JAMA Neurol Charge was funded by the authors. 2016;73:790–794. This is an open access article distributed under the terms of the 5. Karlsson G, Francis G, Koren G, et al. Pregnancy outcomes Creative Commons Attribution-NonCommercial-NoDerivatives Li- in the clinical development program of fingolimod in mul- cense 4.0 (CC BY-NC-ND), which permits downloading and shar- tiple sclerosis. Neurology 2014;82:674–680. ing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from 6. Cavone L, Felici R, Lapucci A, et al. Dysregulation of the journal. sphingosine 1 phosphate receptor-1 (S1P1) signaling and regulatory lymphocyte-dependent immunosuppression in Received January 12, 2017. Accepted in final form May 11, 2017. a model of post-fingolimod MS rebound. Brain Behav Im- – Correspondence to Dr. Novi: [email protected] mun 2015;50:78 86. 7. Voskuhl R. Rebound relapses after ceasing another disease- 1. Miravalle A, Jensen R, Kinkel RP. Immune reconstitution modifying treatment in patients with multiple sclerosis: are inflammatory syndrome in patients with multiple sclerosis there lessons to be learned? JAMA Neurol 2016;73:775–776.

Neurology: Neuroimmunology & Neuroinflammation 3 Clinical/Scientific Notes

Oana M. Dumitrascu, OPSOCLONUS-MYOCLONUS SYNDROME with lymphocytic predominance (91%), absent red MD, MS DURING RITUXIMAB TREATMENT FOR blood cells, normal glucose (55 mg/dL), and mildly Andrew McKeon, MD AUTOIMMUNE AUTONOMIC GANGLIONOPATHY elevated protein (63 mg/dL, normal range 14–45). Leslie Zuniga, MD There were no supernumerary CSF oligoclonal Marie F. Grill, MD bands. Flow cytometry, cytology, and detailed serum Brent P. Goodman, MD Adult-onset opsoclonus-myoclonus syndrome (OMS) and CSF infectious studies for fungal, bacterial, is an autoimmune disorder with paraneoplastic, parain- mycobacterial, spirochetes, and viral pathogens were Neurol Neuroimmunol fectious, or idiopathic etiologies.1 Nonparaneoplastic negative. Comprehensive serum and CSF autoanti- Neuroinflamm 1 2017;4:e376; doi: 10.1212/ cases are generally immunotherapy responsive. body testing demonstrated NMDA-R antibody in NXI.0000000000000376 Although previously described accompanying enceph- the CSF only (at a low titer by cell-based assay alopathy in NMDA receptor (NMDA-R) encephali- [1:2]; indirect immunofluorescence assay using 2 tis, OMS occurring as an isolated CNS finding in an a composite of mouse tissues, including brain, was – NMDA-R antibody positive patient appears to be negative [,1:120]), and a3AChR antibody in serum unusual, particularly among patients with preexisting (titer, 0.69 nmol/L). OMS improved significantly a – 3AChR antibody associated dysautonomia. We within 1 week of initiation of IV methylprednisolone describe a patient in whom an autoimmune CNS treatment (1 g daily for 5 days). The patient was dis- disorder isolated to OMS arose in the context of rit- charged to a skilled nursing facility with plan for a 12- uximab treatment administered for refractory autoim- week course of pulse methylprednisolone: 1 g of mune dysautonomia. methylprednisolone once weekly for 5 more weeks, A 61-year-old woman with a history of prior followed by once every other week dosing for 6 weeks. gastric sleeve procedure to treat morbid obesity pre- Despite improved recognition of OMS in adults sented with profound orthostatic hypotension and and advancements in antibody diagnostics, the path- recurrent syncopal episodes. She was diagnosed with ogenesis of many cases remains incompletely under- severe global autonomic failure involving postgangli- stood.1 Unusual aspects of this report include onic sympathetic sudomotor, cardiovagal, and cardio- NMDA-R autoimmunity occurring in the context vascular adrenergic systems. The serum a3AChR of OMS without NMDA-R-typical encephalopathy antibody was positive (value, 1.30 nmol/L; normal, or neuropsychiatric symptoms and the occurrence of #0.02). CSF studies were not pursued at that time. IV immunoglobulin, mycophenolate mofetil, plasma this clinical-serological constellation in the context of exchange, and symptomatic therapy provided modest rituximab treatment of another autoimmune disor- or transient benefit. Because of suboptimal treatment der. OMS has been previously reported in patients response, rituximab was initiated (1,000 mg, 2 infu- with NMDA-R autoimmunity, where the clinical sions, 2 weeks apart, with diphenhydramine and acet- course was otherwise typical for that disorder, and 2 aminophen as premedication). Within 2 weeks of the encephalitic symptoms were prominent. first infusion, the patient developed progressive ver- The mechanism by which the patient paradoxi- tigo, oscillopsia, gait instability, and diffuse body cally developed OMS after the initiation of rituximab tremulousness. Symptoms worsened after the second is unclear. Rituximab appears to be an effective rituximab infusion. Despite this, autonomic symp- off-label therapy in many cases for autoimmune tomatology had improved, and no behavioral or encephalitis.3 In neuromyelitis optica, relapses of mood changes were encountered. Neurologic exami- optic neuritis and transverse myelitis have been nation 1.5 weeks after the second rituximab infusion reported to occur within 2 weeks of rituximab treat- revealed opsoclonus, myoclonic limb movements, ment.4 It has been hypothesized that rituximab- and truncal ataxia. Contrast-enhanced brain MRI induced B-lymphocyte depletion leads to transient was unrevealing. Chest, abdomen and pelvis CT, increases in serum B-cell activating factor, which in mammography, and whole-body PET/CT did not turn induces a paradoxical upregulation of aquaporin- identify underlying malignancy. CSF evaluation re- 4 autoimmunity.5 Similarly, paradoxical and abrupt vealed 119 nucleated cells/mL (normal less than 5) elevations in serum IgM levels after rituximab therapy

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 initiation have been described in Waldenstrom mac- Disclosure: O.M. Dumitrascu reports no disclosures. A. McKeon has roglobulinemia.6 Biological factors specific to rituxi- patents pending for GFAP and MAP1B as markers of neurological autoimmunity and paraneoplastic disorders; consulted for Grifols, mab in contrast to other novel anti-CD20 MedImmune, and Euroimmun; and received research support from monoclonal antibodies7 may play a role in pathogen- MedImmune. L. Zuniga and M.F. Gill report no disclosures. esis, such as differences in core epitope sequences and B.P. Goodman consulted for Lundbeck. Go to Neurology.org/nn for binding conformations, although precise mechanisms full disclosure forms. The Article Processing Charge was funded by the Mayo Clinic. have yet to be elucidated. It is possible that NMDA-R This is an open access article distributed under the terms of the autoimmunity was triggered by a similar mechanism Creative Commons Attribution-NonCommercial-NoDerivatives Li- in our patient, although a3AChR antibody titer did cense 4.0 (CC BY-NC-ND), which permits downloading and shar- not rise and autonomic symptomatology remained ing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from stable. Coadministration of methylprednisolone will the journal. be considered for future rituximab treatment epochs Received February 3, 2017. Accepted in final form March 23, 2017. in our patient. Similarly, coadministration of “pro- phylactic” corticosteroids with initial rituximab infu- Correspondence to Dr. Dumitrascu: [email protected] sions might be warranted in the management of other 1. Klaas JP, Ahlskog JE, Pittock SJ, et al. Adult-onset autoimmune disorders. opsoclonus-myoclonus syndrome. Arch Neurol 2012;69: Previous studies have demonstrated that rituximab 1598–1607. may paradoxically promote autoantibody production 2. Armangue T, Titulaer MJ, Malaga I, et al. Pediatric anti-N- through a poorly understood mechanism.5 The risk of methyl-D-aspartate receptor encephalitis-clinical analysis new or recurrent immune-mediated neurologic disor- and novel findings in a series of 20 patients. J Pediatr – ders arising as a result of rituximab therapy mandates 2013;162:850 856.e2. 3. Titulaer MJ, McCracken L, Gabilondo I, et al. Treatment close posttreatment monitoring, particularly at the and prognostic factors for long-term outcome in patients present time when the use of B-cell depletion therapy with anti-NMDA receptor encephalitis: an observational for the treatment of autoimmune neurologic diseases cohort study. Lancet Neurol 2013;12:157–165. is increasing. 4. Lindsey JW, Meulmester KM, Brod SA, Nelson F, Wolinsky JS. Variable results after rituximab in neuromyelitis optica. From the Department of Neurology (O.M.D., L.Z., M.F.G., J Neurol Sci 2012;317:103–105. B.P.G.), Mayo Clinic, Phoenix, AZ; and Department of Neurology 5. Nakashima I, Takahashi T, Cree BA, et al. Transient in- (A.M.), Mayo Clinic, Rochester, MN. creases in anti-aquaporin-4 antibody titers following ritux- Author contributions: Oana M. Dumitrascu: drafting/revising of the imab treatment in neuromyelitis optica, in association manuscript, study concept or design, analysis or interpretation of the with elevated serum BAFF levels. J Clin Neurosci 2011; data, accepts responsibility for conduct of research and will give final 18:997–998. approval, acquisition of data, and study supervision. Andrew McKeon: drafting/revising of the manuscript, analysis or interpreta- 6. Treon SP, Branagan AR, Hunter Z, Santos D, Tournhilac tion of the data, and acquisition of data. Leslie Zuniga: drafting/ O, Anderson KC. Paradoxical increases in serum IgM and ’ revising of the manuscript. Marie F. Grill: drafting/revising of the viscosity levels following rituximab in Waldenstrom s mac- manuscript and analysis or interpretation of the data. Brent P. roglobulinemia. Ann Oncol 2004;15:1481–1483. Goodman: drafting/revising of the manuscript, study concept or 7. Klein C, Lammens A, Schafer W, et al. Epitope interactions design, and analysis or interpretation of the data. of monoclonal antibodies targeting CD20 and their rela- Study funding: No targeted funding reported. tionship to functional properties. MAbs 2013;5:22–33.

2 Neurology: Neuroimmunology & Neuroinflammation Clinical/Scientific Notes

Frank Bittner, DO LIPOIC ACID PHARMACOKINETICS AT BASELINE (SD 9.3) years, and the median Expanded Disability Charles Murchison, MS AND 1 YEAR IN SECONDARY PROGRESSIVE MS Status Scale score was 5.5 (range 3.0–8.0). The mean Dennis Koop, PhD baseline Cmax was 14.9 6 11.9 nmol/mL with a non- Dennis Bourdette, MD significant reduction at 1 year (11.3 6 7.3, p 5 0.17, Lipoic acid (LA) is a water- and fat-soluble oral anti- Rebecca Spain, figure, A). At baseline, the largest proportion of sub- oxidant with anti-inflammatory properties. It has MD, MSPH jects (13, 48%) had Cmax values at the 90-minute demonstrated benefits in animal models of MS and draw, whereas at year 1, the largest plurality (9, 41%) has been evaluated for MS relapse prevention and Neurol Neuroimmunol had a Cmax value at the 120-minute draw, although Neuroinflamm neuroprotection. However, there are relatively a few this shift was not significant (p 5 0.47). There was 2017;4:e380; doi: 10.1212/ data regarding LA pharmacokinetics (PK) in elderly NXI.0000000000000380 a nonsignificant reduction in bioavailability at 1 year populations or with use beyond 4 days.1 In addition, (AUC 1407 6 873 nmol/mL vs 1116 6 647 nmol/ studies have used a variety of doses, a wide age range mL, p 5 0.10). Variability as measured by coefficient of subjects, and have measured, at times, specific of variation (CV) was similar at baseline and 1 year enantiomers rather than the more commercially (79.8% vs 64.9%), indicating stability in the PK available racemic form.2 measures, although the within-subject Cmax values at 30 minutes were often discrepant between years Methods. Presented herein are PK results drawn at (158.5% and 179.4%, figure, B). The patients (103, baseline and 1 year in the LA cohort of patients with 114, 144, 147, and 155) terminating early (glomer- secondary progressive MS enrolled in a randomized ulonephritis, MRI intolerance, prostate cancer, placebo-controlled trial of daily oral LA. The study gastrointestinal [GI] intolerance, and renal failure, was approved by the Veterans Affairs Portland Health respectively) did not have observably high Cmax Care System and Oregon Health & Science Univer- levels. sity Institutional Review Boards. Patients arrived after fasting for the prior 10 hours, and a predose sample Discussion. Overall, patients maintained peak was taken. Patients ate a meal immediately followed serum levels of daily oral LA, although there were by 1,200 mg racemic LA (Pure Encapsulations, nonsignificant reductions toward lower and later Sudbury, MA). Blood draws occurred at 30, 60, 90, absorptions at 1 year. Cmax values occurred later 120, and 240 minutes after dose. Blood was allowed (between 90 and 120 minutes) than a previous PK to clot at room temperature; serum was separated study of LA using the same dosing regimen (between by centrifugation and stored at 280°C until batch 60and90minutes).3 Because of limited clearance analysis by mass spectrometry.3 data, the analysis was unable to calculate many com- Noncompartmental analysis determined pertinent mon, tail-based noncompartmental analysis param- PK parameters, including peak concentration eters, including half-life. Although the mean Cmax (Cmax), time at peak concentration (Tmax), and values were similar between baseline and 1 year, observed bioavailability based on area under the curve visual observation demonstrates high between- (AUC) using common pharmacodynamics calcula- subject variability for the same year and within- tions. Baseline and 1-year differences were assessed subject variability between years based on the high using mixed models to account for serial correlation coefficients of variation (CV .65%). A review of in the repeated measures and accommodate subjects apparent outliers (115, 134, 137, and 149) did not with missing data at 1 year. reveal underlying differences (e.g., age, weight, and Results. Fifty-four patients were randomized in the concomitant medications), nor were their mean parent trial, and of the 28 assigned to LA, 27 took brain atrophy rates different from the larger cohort. at least 1 dose of LA and were included in PK analysis. Breithaupt-Grögler et al. (1999) also noted high Patients demonstrated 87% compliance by pill between-individual variability in Cmax values of LA counts. The average age of the LA cohort was 57.9 (99%and96%ofthemeasuredRandSLAenan- (SD 6.7) years, 59% were women, and 96% were tiomers at the highest dose of 600 mg of racemic Caucasian. The average disease duration was 30.9 LA). Reasons for between- and within-subject

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Figure Pharmacokinetic concentration vs time plots

(A) LA concentration at 6 time points over 120 minutes at baseline (n 5 27) and 1 year (n 5 22). Shown are mean values with SD bars. (B) Individual traces of baseline and 1-year mean LA peak concentrations. Variability measured by the mean coefficient of variation across the pharmacokinetic trace was similar at the 2 time points (79.8% vs 64.9%, respectively) with the highest variability found at 30 minutes (158.5% and 179.4%, respectively). LA 5 lipoic acid. variable absorptions may be due to an elderly pop- it may relate to intrinsic properties of LA or its ulation with erratic GI absorption, reduced hepatic delivery system.4–6 Yet unknown is if the PK vari- perfusion, or drug-drug interactions. Alternatively, ability and rapid clearance of LA impacts its

2 Neurology: Neuroimmunology & Neuroinflammation therapeutic efficacy or has dosing implications for This is an open access article distributed under the terms of the clinical trials or clinical use. Further development of Creative Commons Attribution-NonCommercial-NoDerivatives Li- cense 4.0 (CC BY-NC-ND), which permits downloading and shar- LA may depend on improving its bioavailability and ing the work provided it is properly cited. The work cannot be tolerability. These PK data represent the longest changed in any way or used commercially without permission from durationuseofLAinanMS-specific population. the journal.

From the VA Portland Health Care System (F.B., C.M., D.B., R.S.), Received March 10, 2017. Accepted in final form May 21, 2017. OR; and Oregon Health & Science University (F.B., C.M., D.K., D. B., R.S.), Portland. Correspondence to Dr. Bittner: [email protected] Author contributions: Frank Bittner: manuscript writing. Charles Murchison: statistical analysis and manuscript editing. Dennis Koop: data collection and manuscript editing. Dennis Bourdette: data 1. Teichert J, Hermann R, Ruus P, Preiss R. Plasma kinetics, interpretation. Rebecca Spain: data collection, data interpretation, metabolism, and urinary excretion of alpha-lipoic acid fol- and manuscript editing. lowing oral administration in healthy volunteers. J Clin Acknowledgment: Pure Encapsulations, Sudbury, MA, provided the Pharmacol 2003;43:1257–1267. lipoic acid and placebo. 2. Hermann R, Mungo J, Cnota P, Ziegler D. Enantiomer- Study funding: Department of Veterans Affairs (B7493-W, selective pharmacokinetics, oral bioavailability, and sex R. Spain), National Institutes of Health (UL1TR000128). effects of various alpha-lipoic acid dosage forms. Clin Phar- Disclosure: F. Bittner received travel funding from the National Mul- macol 2014;6:195–204. tiple Sclerosis Society. C. Murchison received research support from NINDS. D. Koop served on the editorial board for Drug Metabolism 3. Yadav V, Marracci G, Munar M, et al. Pharmacokinetic and Disposition. D. Bourdette received travel funding from the National study of lipoic acid in multiple sclerosis: comparing mice Multiple Sclerosis Society, Consortium of MS Centers, and Paralyzed and human pharmacokinetic parameters. Mult Scler 2010; Veterans of America; is on the editorial board for Neurology;holds 16:387–397. a patent for the treatment of multiple sclerosis with cyclic peptide deriv- 4. Breithaupt-Grögler K, Niebch G, Schneider E, et al. Dose- atives of cyclosporin; has a patent pending for thyromimetic drugs for proportionality of oral thioctic acid–coincidence of assess- stimulating remyelination in multiple sclerosis; consulted for Magellan ments via pooled plasma and individual data. Eur J Pharm Health, Best Doctors, inc; and received research support from the Sci 1999;8:57–65. NationalMSCenter.R.Spainreceived research support from Biogen, 5. Klotz U. Pharmacokinetics and drug metabolism in the Department of Veterans Affairs, Oregon Clinical and Translational elderly. Drug Metab Rev 2009;41:67–76. Research Institute, VA Portland Health Care System, Oregon Health & Science University, and National MS Society, Conrad Hilton 6. Phua L, New L, Goh C, Neo A, Browne E, Chan E. Foundation, Medical Research Foundation of Oregon, Race to Erase Investigation of the drug-drug interaction between alpha- MS. Go to Neurology.org/nn for full disclosure forms. The Article lipoic acid and valproate via mitochondrial beta-oxidation. Processing Charge was funded by the authors. Pharm Res 2008;25:2639–2649.

Neurology: Neuroimmunology & Neuroinflammation 3 Clinical/Scientific Notes

Salwa Kamourieh, MBBS, NATALIZUMAB GRANULE CELL not traditionally used, a 5-day trial of G-CSF was PhD NEURONOPATHY: FDG-PET IN DIAGNOSIS AND prescribed. Over the next 6 weeks, she showed pro- Kohilan Gananandan, IMMUNE RECONSTITUTION WITH G-CSF gressive improvement in speech and mobility, and her MBBS EDSS improved to 7.0. A repeat CSF in January Joel Raffel, BMBCh Granulecellneuronopathy(GCN)isararepresen- 2016 showed that the viral load had dropped to Richard Nicholas, MBBS, tation of JC virus (JCV) infection targeting the 250 copies/mL JCV-DNA. PhD cerebellum.1 To date, 2 cases have been described In February 2016, she experienced severe nausea. in patients prescribed natalizumab for MS.2,3 Recent Her MRI showed high signal in the brainstem, in par- Neurol Neuroimmunol ticular the pons (figure, B.b red arrow), felt to be Neuroinflamm case reports have acknowledged the importance of 2017;4:e384; doi: 10.1212/ controlled reconstitution of host immunity in treat- consistent with IRIS. She declined steroids feeling NXI.0000000000000384 ing JCV-related neurologic disorders.4 In particular, this treatment was not warranted, given her overall granulocyte-colony stimulating factor (G-CSF) has clinical improvement. To date, her most recent imag- emerged as a potential treatment.5 ing demonstrated a stable appearance and her latest Here, we report a case of natalizumab-associated EDSS is 6.5 (figure, A). GCN where diagnosis was aided by the novel use of Discussion. This is the first case to describe the use FDG-PET, and immune reconstitution driven by of FDG-PET to aid diagnosis and the use of G-CSF G-CSF lead to disease stabilization and subsequent to promote immune reconstitution in an MS patient improvement. with GCN. Although our understanding of PML after natalizumab is improving, our understanding of Case report: clinical course. A 31-year-old Cau- GCN after natalizumab is limited. casian female diagnosed in 1990 with relapsing- Difficulty in diagnosing early GCN is highlighted remitting MS on natalizumab since 2008 was here, where progressive cerebellar atrophy was identi- admitted twice between July and September 2015, fied retrospectively, long after JCV-DNA was con- with progressive unsteadiness, dysarthria, and diplo- firmed in the CSF. The presence of early cerebellar pia. Her serum anti-JCV antibody index was 2.41. symptoms in patients treated with natalizumab re- Following concerns about possible progressive mul- quires early and regular imaging for cerebellar atrophy, tifocal leukoencephalopathy (PML), her natalizumab in particular progression of atrophy,6 with or without was stopped in July 2015 (figure, A). She was treated lumbar punctures for JCV-DNA detection.7 The use with IV methylprednisolone, and the second course of FDG-PET here is novel, demonstrating hypome- was followed up with oral prednisolone, to reduce her tabolism within both hemispheres of the cerebellum risk of immune reconstitution inflammatory syndrome and enabling clear anatomical localization of the prob- (IRIS). Despite unchanged imaging (figure, B), in lem. Hypometabolism is an expected finding given the September 2015, a lumbar puncture found JCV-DNA pathologic findings of neuronal cell loss in GCN, viral load measured at 5,000 copies/mL. whereas in IRIS one may have expected more activity Her clinical condition deteriorated and she resulting from influx of immune cells. Limitations of became bedbound, and, in November 2015, her this include the delay in performing the scan and the Expanded Disability Status Scale (EDSS) measured absence of serial imaging. FDG-PET could potentially 8.0 on admission. A repeat CSF sample measured be useful if used earlier; however, further studies into a viral load of 17,000 copies/mL of JCV-DNA. the utility of FDG-PET in this context are needed. Repeat MRI did not identify features consistent with This patient was not a candidate for plasmapheresis PML but did show progressive cerebellar volume loss becauseshehadbeenoffnatalizumabfor3monthsand (figure, B.b). A fluorodeoxyglucose-PET (FDG-PET) it would have been cleared from her system. In addition, demonstrated patchy hypometabolism in the cerebel- plasma exchange has been shown to be ineffective in lum (figure, C). Based on these results, a diagnosis of natalizumab-associated PML cases.8 Faced with progres- GCN without PML was made. Mirtazapine was sive clinical deterioration and rising JCV-DNA in the started, and prednisolone was stopped. Although CSF despite steroids, it was decided to start G-CSF,

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 Figure (A) Diagrammatic representation of disease course, EDSS, JCV CSF viral load, and treatments

(A) The disease was managed with interferon beta-1a (Rebif®) (2007–2009), but this was changed to monthly infusions of natalizumab in January 2009 because of disease breakthrough while on treatment. On starting natalizumab, her EDSS was 2.5; despite testing positive for anti-JCV antibody, her treatment was continued until July 2015, with a stable EDSS 2.0 and no further relapses. Her EDSS went up to 6.0 in July and to 8.0 in November. After treatment with G-CSF, her EDSS decreased to 6.5. (B) Diagram showing sequential MRI slices. Axial slices of T2 FLAIR sequences. Upper panel (B.a) dem- onstrates the unchanged multiple foci of abnormal signal in the cerebral white matter. Lower panel (B.b) demonstrates progressive cerebellar volume loss in particular between September 2015 and November 2015. The red arrow identifies an area of high signal in the pons. (C) FDG-PET. Brain PET FDG identifying small patches of hypometabolism (blue arrows) in the left inferior cerebellar cortex and right posterior cerebellar cortex, performed on December 31, 2015. EDSS 5 Expanded Disability Status Scale; FDG-PET 5 fluorodeoxyglucose PET; FLAIR 5 Fluid attenuation inversion recovery; G-CSF 5 granulocyte-colony stimulating factor; JCV 5 JC virus.

a glycoprotein, which primarily stimulates the produc- IRIS in 16/18 patients.5 This is the first case to describe tion, mobilization, and survival of neutrophils, to initiate the use of G-CSF in the treatment of GCN but similar immune reconstitution in December 2016. Stefoski to its prior use, our patient developed IRIS approxi- highlighted the clinical benefit of G-CSF in PML in mately 6 months after natalizumab and 2 months after 8/16 patients, with the desired response of immune- G-CSF. This was managed conservatively because of no boosting leukocytosis and lymphocytosis resulting in significant clinical worsening.

2 Neurology: Neuroimmunology & Neuroinflammation This case highlights that the use of a novel 1. Tan CS, Koralnik IJ. Progressive multifocal leukoence- diagnostic tool in GCN and hypometabolism on phalopathy and other disorders caused by JC virus: clin- FDG-PET in the cerebellum could be used to sup- ical features and pathogenesis. Lancet Neurol 2010;9: 425–437. port the early diagnosis of GCN. We also demon- 2. Agnihotri SP, Dang X, Carter JL, et al. JCV GCN in strate as had been shown in preliminary case reports a natalizumab-treated MS patient is associated with in PML that the use of G-CSF in these patients mutations of the VP1 capsid gene. Neurology 2014;83: may be helpful in GCN. 727–732. 3. Schippling S, Kempf C, Buchele F, et al. JC virus granule From the Department of Neurology (S.K., J.R., R.N.), Imperial Col- cell neuronopathy and GCN-IRIS under natalizumab treat- lege NHS Trust, London; and Department of Medicine (K.G.), ment. Ann Neurol 2013;74:622–626. Ealing Hospital, Southall, United Kingdom. 4. Brew BJ, Davies NW, Cinque P, Clifford DB, Nath A. Author contributions: S. Kamourieh, K. Gananandan, J. Raffel, and Progressive multifocal leukoencephalopathy and other — R. Nicholas write up of clinical note. forms of JC virus disease. Nat Rev Neurol 2010;6: Study funding: No targeted funding reported. 667–679. Disclosure: S. Kamourieh and K. Gananandan report no disclosures. 5. Stefoski D, Balabanov R, Ko M, Strong J, Javed A. Les- J. Raffel received research support from GE Healthcare and National sons learned from eighteen non-fatal outcomes in Multiple Sclerosis Society. R. Nicholas served on the scientific advisory natalizumab-associated PML: an evolving therapeutic board for Biogen, Roche, Novartis, and Genzye; received travel fund- approach in settings of varied presentations. Mult Scler J ing and/or speaker honoraria from Biogen; and received research 2015;21:586–587. support from Multiple Sclerosis Society. Go to Neurology.org/nn for full disclosure forms. The Article Processing Charge was funded by 6. Wijburg MT, Siepman D, van Eijk JJ, Killestein J, Wattjes Imperial College London. MP. Concomitant granule cell neuronopathy in patients This is an open access article distributed under the terms of the with natalizumab-associated PML. J Neurol 2016;263: – Creative Commons Attribution-NonCommercial-NoDerivatives Li- 649 656. cense 4.0 (CC BY-NC-ND), which permits downloading and shar- 7. Wattjes MP, Vennegoor A, Mostert J, et al. Diagnosis of ing the work provided it is properly cited. The work cannot be asymptomatic natalizumab-associated PML: are we between changed in any way or used commercially without permission from a rock and a hard place? J Neurol 2014;261:1139–1143. the journal. 8. Landi D, De Rossi N, Zagaglia S, et al; Italian PML study Received February 24, 2017. Accepted in final form May 30, 2017. group. No evidence of beneficial effects of plasmapheresis in natalizumab-associated PML. Neurology 2017;88: Correspondence to Dr. Kamourieh: [email protected] 1144–1152.

Neurology: Neuroimmunology & Neuroinflammation 3 Clinical/Scientific Notes

Rannveig Viste, MSc LONG-TERM IMPROVEMENT AFTER COMBINED hallucinations emerged, but no microsleeps were Joseph Soosai, MD IMMUNOMODULATION IN EARLY POST-H1N1 reported; he took only 1 daily nap. The CSF Truls Vikin, MD VACCINATION NARCOLEPSY hypocretin-1 levels had dropped to undetectable Per Medbøe Thorsby, levels (,40 pg/mL). From October 2014 to Janu- – MD, PhD We previously described the possible clinical effects ary 2015, his status was 3 4 naps/day, cataplexy Kristian Bernhard Nilsen, of early monotherapeutic IV-immunomodulation and hypnagogic hallucinations several times/week. MD, PhD (IVIg) treatment in sporadic1 but not in post- Follow-up MSLT mean sleep latencies in January Stine Knudsen, MD, PhD H1N1 vaccination narcolepsy type 1 (NT1).2 We 2014 and June 2015 were normal (11.5 minutes report here an early post-H1N1 vaccination NT1 and 12.5 minutes, respectively). Neurol Neuroimmunol case treated with combined immunomodulation of In late March 2015, another, otherwise healthy, Neuroinflamm – 2017;4:e389; doi: 10.1212/ IVIg and methylprednisolone, and a comparable 6-year-old boy (child 2) developed 2 4 sleep NXI.0000000000000389 sporadic NT1 case. attacks/day. His nocturnal sleep was disrupted by awakenings, nightmares, and dream enactment, and cataplexy (tongue protrusions and unsteady Case report. In late 2009, a 2.5-year-old boy (child 1) gait) became apparent in May 2015. Primo July was H1N1 vaccinated with H1N1 pandemic 2015, the ESS score was 17/24, and he presented influenza vaccine (Pandemrix; GlaxoSmithKline, extensive spontaneous cataplexy, additionally exac- London, United Kingdom). He had a history of erbated by joy/excitement. MSLT mean sleep asthma/allergies and sleep talking/sleepwalking. Me- latency was 2.1 minutes with SOREMs in 4/5 dio August 2013, he developed severe sleepiness, took naps. CSF hypocretin-1 levels were ,40 pg/mL. 2–4 naps/day, and experienced episodes of sleep A similar treatment regime as for child 1 was initi- paralysis. We examined him primo September 2013 ated. During T1 methylprednisolone treatment, when the baseline Epworth Sleepiness Scale (ESS) the ESS score dropped transiently to 11/24, and score was 14/24 and multiple sleep latency test the frequency of cataplexy decreased from 8 to 6 (MSLT) showed a mean sleep latency of 5 minutes attacks/day. After T1, the ESS score returned to with sleep onset REMs (SOREMs) in 4/4 nap 16/24, and cataplexy gradually decreased to 3 at- opportunities. CSF hypocretin-1 levels were low tacks/day, remaining at that frequency until follow- (77 pg/mL; normal values $200 pg/mL). up. Medio October, the follow-up ESS score was In late September 2013, cataplexy (tongue pro- 16/24, and MSLT mean sleep latency was 6 mi- trusion, facial muscles hypotonia, and head drop) nuteswithSOREMsin4/4naps.Hestillhaddis- triggered by laughter/emotional stress emerged rupted night sleep, and cataplexy had increased to and rapidly escalated to several partial/fulminant at- 5–7 attacks/day. The CSF hypocretin-1 levels were tacks/day. In the beginning of October 2013, hav- unchanged (table). ing obtained informed consent, we initiated treatment of IVIg 1 g/kg/day for 2 consecutive Discussion. The present post-H1N1–vaccinated days, followed by methylprednisolone 20 mg/kg/ NT1 case treated early after disease onset with com- day for 4 days, administered 3 times at monthly in- bined IVIg and methylprednisolone is notable for tervals (T1, T2, and T3). During T1, T2, and T3 several reasons. First, the transient abrupt clinical methylprednisolone treatment days, cataplexy com- improvement during the methylprednisolone infu- pletely disappeared: he did activities such as watch- sion, a drug thought mainly to target cellular immune ing cartoons and making jokes, without muscle mechanisms, supports the recent indications of anti- weakness. However, cataplexy gradually reappeared gen presentation to T cells as central factors in nar- 1–2 weeks after treatment. Likewise, the ESS score colepsy pathophysiology.3 Second, although his CSF normalized from 14/24 to 3/24 during T1 methyl- hypocretin-1 levels subsequently dropped to unde- prednisolone treatment, from 8/24 to 2/24 during tectable levels, there was long-term clinical T2 IVIg and methylprednisolone, and from 14–16/ improvement, as the objective sleep latencies were 24 to 7/24 after T3. After T3, hypnagogic normal, cataplexy frequency was reduced, and he

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Table Demographic, clinical, and paraclinical data

Child 1 Child 2

Sex/age Male/6 y 6 mo Male/6 y 2 mo

Ethnicity White White

H1N1 vaccinated (Pandemrix) Autumn 2009 No

Weight gain (yes-no) No Yes, 5 kg

Onset EDS (month/year) Mid August 2013 Late March 2015

Onset cataplexy (month/year) Late September 2013 May 2015

Follow-up

Baseline Baseline Follow-up Immunomodulation period October 2013 January 2014 June 2015 July 2015 October 2015

ESS (score/24) 14/24 7/24 12/24 17/24 16/24

Cataplexy (episodes per day) Multiple 6135–7

Hypnagogic hallucinations (yes-no) No Yes No No No

Sleep paralysis (yes-no) Yes No No No No

Disturbed night sleep (yes-no) Yes Yes Yes Yes Yes

Dream enactment (yes-no) Yes Yes Yes (but less) Yes Unknowna

Medication (yes-no) No No No No No

PSG total sleep time (minutes) 603 674 600 560 NAb

PSG sleep latency, min 6 1 12.1 1.9 NAb

PSG SOREM (yes-no) Yes No No No NAb

MSLT MSL, min 5.0 11.5 12.5 2.1 6.0

MSLT SOREMs (number/naps) 4/4 3/4 5/5 4/5 4/4

CSF hypocretin-1 concentration (pg/mL; low ,150 pg/mL)c 77 ,40 Not measured ,40 ,40

Abbreviations: EDS 5 excessive daytime sleepiness; ESS 5 Epworth Sleepiness Scale; MSL 5 mean sleep latency; MSLT 5 multiple sleep latency test; NA 5 not available; PSG 5 polysomnography; SOREM 5 sleep onset REM. a A precise description of dream enactment symptoms was not specified at the time of follow-up examination. b Because of technical reasons, data from PSG were lost, but the parent who coslept reported that he fell asleep after 5–10 minutes and slept from 19:55 to 06:40. c Baseline and follow-up hypocretin-1 measurements were analyzed in the same radioimmunoassay (Phoenix) kit to avoid interassay variability.

managed school without medication at 18 months As child 1, but not child 2, improved clinically and of follow-up. initially had a low but still detectable CSF Apart from a single very early NT1 case, previous hypocretin-1 level, we speculate that early (enough) monoimmunotherapy and a single combined immu- combined humoral and cellular immunomodula- nomodulation attempt in sporadic NT1 produced tion can result in long-term disease improvement mild/moderate subjective, but not objective improve- in human narcolepsy. ments.1 Based on the strong HLA-DQB1*0602 asso- 4 3 From the Norwegian Centre of Expertise for Neurodevelopmental ciation, possible antigen presentation to T cells, and Disorders and Hypersomnias (NevSom), Department of Rare Disor- additional immunogenic polymorphisms,5 narcolepsy ders (R.V., S.K.), Department of Neurology, Section for Clinical is strongly believed to be an autoimmune disease. In Neurophysiology (R.V., K.B.N.), Department of Child Neurology animal models as well as in a human case,6 hypocretin (J.S., T.V.), Oslo University Hospital, Ullevål, Norway; Hormone Laboratory, Department of Medical Biochemistry (P.M.T.), Oslo deficiency occurs around the time of clinical disease University Hospital, Aker, Norway; Department of Neuromedicine onset. Although it is not known exactly what the and Movement Science (K.B.N.), Norwegian University of Science potential immunomodulation therapeutic window and Technology, Trondheim, Norway; and Medical Faculty (R.V.), University of Oslo, Norway. is, early chronical combined immunomodulation Author contributions: R. Viste: handled the patients, analyzed and (including steroids) in genetic narcoleptic dogs interpreted the investigation data, and wrote the manuscript. J. Soosai resulted in long-term milder disease. The direct and T. Vikin: designed and coordinated the study and handled the pharmacologic effect on narcoleptic symptoms patients. P.M. Thorsby: designed the study, analyzed CSF Hcrt-1/typed (for example direct anticataplectic effect) was not HLA, and revised the manuscript. K.B. Nilsen: designed the study, analyzed and interpreted the investigation data, and revised the man- observed in these dogs, so combined immunomo- uscript. S. Knudsen: designed the study, handled the patients, analyzed dulation was the probable cause of improvement.7 and interpreted the investigation data, and wrote the manuscript.

2 Neurology: Neuroimmunology & Neuroinflammation Study funding: No targeted funding reported. 2. Knudsen S, Biering-Sorensen B, Kornum BR, et al. Early Disclosure:R.Viste,J.Soosai,T.Vikin,P.M.Thorsby,andK.B.Nilsen IVIg treatment has no effect on post-H1N1 narcolepsy report no disclosures; S. Knudsen has been an expert consultant for the phenotype or hypocretin deficiency. Neurology 2012;79: Norwegian state and received research support from the Norwegian Min- 102–103. istry of Health and Care Services. Go to Neurology.org/nn for full disclo- 3. Faraco J, Lin L, Kornum BR, et al. ImmunoChip study sure forms. The Article Processing Charge was funded by Norwegian implicates antigen presentation to T cells in narcolepsy. Centre of Expertise for Neurodevelopmental Disorders and Hypersomnias PLoS Genet 2013;9:e1003270. (NevSom). 4. Mignot E, Lin L, Rogers W, et al. Complex HLA-DR and This is an open access article distributed under the terms of the -DQ interactions confer risk of narcolepsy-cataplexy in Creative Commons Attribution-NonCommercial-NoDerivatives Li- three ethnic groups. Am J Hum Genet 2001;68:686–699. cense 4.0 (CC BY-NC-ND), which permits downloading and shar- ing the work provided it is properly cited. The work cannot be 5. Kornum BR, Kawashima M, Faraco J, et al. Common var- changed in any way or used commercially without permission from iants in P2RY11 are associated with narcolepsy. Nat Genet – the journal. 2011;43:66 71. Received April 26, 2017. Accepted in final form June 13, 2017. 6. Savvidou A, Knudsen S, Olsson-Engman M, Gammeltoft S, Jennum P, Palm L. Hypocretin deficiency develops during Correspondence to R. Viste: [email protected] onset of human narcolepsy with cataplexy. Sleep 2013;36: 147–148. 1. KnudsenS,MikkelsenJD,GammeltoftS,JennumP.Intrave- 7. Boehmer LN, Wu MF, John J, Siegel JM. Treatment with nous immunoglobulin treatment and screening for hypocretin immunosuppressive and anti-inflammatory agents delays neuron-specific autoantibodies in recent onset childhood nar- onset of canine genetic narcolepsy and reduces symptom colepsy with cataplexy. Neuropediatrics 2010;41:217–222. severity. Exp Neurol 2004;188:292–299.

Neurology: Neuroimmunology & Neuroinflammation 3 Clinical/Scientific Notes

Andreia Costa, MD, POSTMALARIA NEUROLOGIC SYNDROME a nonpainful, nontender mass in the posterior cervical MSc* ASSOCIATED WITH NEUREXIN-3a ANTIBODIES region were noted. He was disoriented in time and André Silva-Pinto, MD, space, had severe attention deficit, acalculia, and vi- MSc* suospatial difficulties. Blood laboratory tests showed Joana Alves, MD, MSc A 57-year-old Caucasian man with no clinical history mild hypochromic anemia, with no further abnor- Nélia Neves, MD, MSc of interest returned to Portugal after a long stay in malities. No Plasmodium parasites were found in Eugenia Martínez- rural Angola. Fifteen days before returning to Portugal, the thin blood smear. Brain CT was normal, and Hernández, MD, PhD he had visited a tropical forest region in the North of the CSF study showed 159 cells (91% mononuclear), Pedro Abreu, MD Angola. Five days after returning, he experienced protein 2.12 g/dL, normal glucose CSF/serum ratio, António Sarmento, MD, fever, asthenia, dizziness, and myalgia. He was diag- and normal IgG index. EEG was consistent with PhD nosed with Plasmodium falciparum malaria (8% para- a severe subcortical encephalopathy with no epileptic sitemia), with no severity criteria according to the activity. Brain MRI showed a mild increase of T2/ Neurol Neuroimmunol World Health Organization guidelines.1 He was suc- fluid-attenuated inversion recovery signal in the Neuroinflamm 2017;4:e392; doi: 10.1212/ cessfully treated with IV quinine-doxycycline and dis- caudate-capsule-lenticulate regions with no gadoli- NXI.0000000000000392 charged home 11 days after fully recovered. nium enhancement (figure). Extensive serum and Two weeks later, the patient developed fever, apa- CSF studies were negative for active infectious (table) thy, somnolence, sleep-awake cycle disruption, con- or systemic autoimmune causes (negative antinuclear fusion, and altered behavior, progressing for 7 days. antibodies, anti–neutrophil cytoplasmic antibodies, On examination, he was febrile but hemodynamically anti–double-strand DNA, anti–extractable nuclear stable. A mild macular rash in the trunk and antigens, rheumatoid factor, anti–amphiphysin 1,

Figure Brain MRI

Mild increase of T2/fluid-attenuated inversion recovery (FLAIR) signal in the caudate-capsule-lenticulate regions with no ga- dolinium enhancement. (A) T1, (B) T1 with gadolinium, (C) T2 FLAIR, (D) diffusion-weighted image, (E) apparent diffusion coef- ficient, and (F) gradient echo.

Neurology.org/nn Copyright © 2017 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology 1 neuropil of rat brain that was different from other Table Infectious studies previously known antibodies, and by specific cell-

Cervical lymph based assays, which confirmed positive serum and Test IgM IgG Blood CSF node CSF reactivity with human embryonic kidney 293 Culture (HEK293)-expressing neurexin-3a, and negative Bacterial 2(32) 22 reactivity with HEK293 expressing the NMDA,

Fungal 2 AMPA, GABA-A, GABA-B, mGluR1, and mGluR5 receptors, or LGI1, Caspr2, DPPX, and IgLON5 Serology proteins. Borrelia afzelii and garinii 22 A new form of autoimmune encephalitis associ- 11 CMV ated with neurexin-3a antibodies was recently EBV 21 described in 5 patients.2 One of them may have HSV1 21 had a history of malaria a few months before devel-

HSV2 11 oping autoimmune encephalitis (patient 1 of Gresa- 2 Mycoplasma species 22 Arribas et al. ). This patient, a 23-year-old Caucasian woman, presented with a 2-week history of worsen- Rickettsia conorii 2 ing headache, somnolence, disorientation, confu- 2 TPPA sion, and recent memory problems. Two months Trypanosoma brucei gambiensea 2 before, she had returned from a 6-month trip to Wright reaction 2 Tanzania. During her stay in a rural area in Tanzania, Nucleic acid amplification test she had 2 episodes of fever which had been treated as

CMV 2 malaria (despite no microbiologic confirmation). On admission, she was disoriented in time and Enterovirus 2 space, with emotional lability, short-term memory HSV 2 deficits, and confabulations. Blood smear test was Tropheryma whipplei 2 negative for parasites. Transient leukopenia and Trypanosoma speciesa 222 thrombocytopenia were detected in blood tests. VZV 2 CSF showed 20 cells with normal protein and glu-

West Nile virus 2 cose; brain MRI was normal; and EEG showed slow

Direct identification activity on left posterior temporal areas. Extensive microbiologic studies were negative in blood and Plasmodium species (thin blood smear) 2(33) CSF. She was treated with IV methylprednisolone a 222 Trypanosoma species with improvement. Abbreviations: CMV 5 cytomegalovirus; EBV 5 Epstein-Barr virus; HSV 5 herpes simplex We report here another patient whose neurologic virus; TPPA 5 Treponema pallidum particle agglutination enzyme immunoassay; VZV 5 symptoms were confirmed preceded by malaria and varicella zoster virus. who had a rapid improvement after steroid treatment. a Performed in the national reference laboratory for Trypanosoma species (Instituto de Higiene e Medicina Tropical, Universidade Nova de Lisboa, Lisboa, Portugal). PMNS was first described, mainly in adults after recovery of severe falciparum malaria, in 1996 in Vietnam.3 There are some previous descriptions of anti-CV2/CRMP5, anti-Ma2, anti-Ri, anti-Yo, and delayed cerebellar ataxia following malaria in India anti-Hu antibodies in serum). and Sri Lanka.4,5 The clinical picture of PMNS is Neurologic symptoms progressed for the next 8 characterized by neuropsychiatric manifestations, days, and postmalaria neurologic syndrome (PMNS) including confusion, altered behavior, decreased con- was diagnosed. A 5-day course of 1 g IV methylpred- sciousness, hallucinations, and seizures. The neuro- nisolone resulted in a significant clinical improve- logic syndrome is self-limited, and no specific ment. A follow-up CSF study showed 34 cells, treatment is required, although in severe cases steroids protein 1.7 g/dL, normal glucose CSF/serum ratio, may be beneficial.6 Immune-mediated mechanisms and no oligoclonal bands. The patient was discharged have been postulated in this postinfectious diffuse on day 15, asymptomatic. Patient samples were encephalopathy, but no autoantibodies had been pre- sent to a neuroimmunology referral center (Professor viously reported. Josep Dalmau Laboratory, Neuroimmunology, The case reported here suggests that encephalitis IDIBAPS, Barcelona, Spain), which showed the pres- associated with neurexin-3a antibodies can poten- ence of neurexin-3a antibodies.2 The presence of cell tially manifest as postinfectious encephalitis. Recog- surface/synaptic neuronal antibodies was determined nition of this disorder is important because it by rat brain immunohistochemistry, which showed responds to immunotherapy. Similar investigations a distinctive pattern of serum and CSF reactivity with are worth to consider in patients with malaria and

2 Neurology: Neuroimmunology & Neuroinflammation especially in patients with PMNS to further elucidate disclosures. Go to Neurology.org/nn for full disclosure forms. The Article the relationship between these entities. Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative * These authors contributed equally to the manuscript. Commons Attribution-NonCommercial-NoDerivatives License 4.0 From the Department of Neurology (A.C., P.A.), Department of (CC BY-NC-ND), which permits downloading and sharing the work Infectious Diseases (A.S.-P., J.A., N.N., A.S.), Centro Hospitalar provided it is properly cited. The work cannot be changed in any way or São João, Porto, Portugal; Department of Neuroscience and Mental used commercially without permission from the journal. Health (A.C., P.A.), Faculty of Medicine of University of Porto, Received April 7, 2017. Accepted in final form July 3, 2017. Portugal; Nephrology and Infectious Diseases R&D Group (A.S.-P., J.A., N.N., A.S.), Health Investigation and Innovation Institute (I3S), Correspondence to Dr. Costa: [email protected] University of Porto, Portugal; and Neuroimmunology Program (E.M.-H.), Institut d’Investigacions Biomèdiques August Pi i Sunyer (IDIBAPS), Barcelona, Spain. 1. Guidelines for the Treatment of Malaria, 3rd ed. World Author contributions: A. Costa and A. Silva-Pinto: clinical manage- Health Organization: Geneva; 2015. ment of the patient, description of the clinical case, and elaboration of 2. Gresa-Arribas N, Planaguma J, Petit-Pedrol M, et al. Human the manuscript. J. Alves: clinical management of the patient, espe- neurexin-3alpha antibodies associate with encephalitis and cially regarding Trypanosoma exclusion, and revision of the manu- alter synapse development. Neurology 2016;86:2235–2242. script. N. Neves: clinical management of the patient and revision of 3. Nguyen TH, Day NP, Ly VC, et al. Post-malaria neuro- the manuscript. E. Martínez-Hernández: laboratorial workout, – description of the laboratorial techniques, and revision of the manu- logical syndrome. Lancet 1996;348:917 921. script. P. Abreu and A. Sarmento: study concept and critical revision 4. Senanayake N, de Silva HJ. Delayed cerebellar ataxia com- of the manuscript for intellectual content. plicating falciparum malaria: a clinical study of 74 patients. – Acknowledgment: The authors thank Professor Josep Dalmau for the J Neurol 1994;241:456 459. scientific contribution to the manuscript. 5. Kochar DK, Kumawat BL, Kochar SK, Bajiya HN, Maurya – Study funding: This study was not funded. RK. Delayed cerebellar ataxia a complication of Plasmo- Disclosure: A. Costa, A. Silva-Pinto, J. Alves, and N. Neves report no dium falciparum malaria. J Assoc Physicians India 1996;44: – disclosures. E. Martinez-Hernandez received research support from In- 686 688. stituto de Salud Carlos III. P. Abreu received speaker honoraria from 6. Markley JD, Edmond MB. Post-malaria neurological syn- Biogen, Bayer, Novartis, Sanofi Genzyme, Merck Serono, and Boehringer drome: a case report and review of the literature. J Trav Ingelheim and consulted for Biogen and Bayer. A. Sarmento reports no Med 2009;16:424–430.

Neurology: Neuroimmunology & Neuroinflammation 3 CORRECTION

Both cladribine and alemtuzumab may effect MS via B-cell depletion In the article “Both cladribine and alemtuzumab may effect MS via B-cell depletion” by D. Baker et al.,1 the title should read “Both cladribine and alemtuzumab may affect MS via B-cell depletion.” The authors regret the error. Additionally, in the key for figure 2 the red diamond should indicate “CLAD 3.5 mg/kg” rather than “CLAD 6.5 mg/kg” as originally published. The editorial staff regret the error.

REFERENCE 1. Baker D, Herrod SS, Alvarez-Gonzalez C, Zalewski L, Albor C, Schmierer K. Both cladribine and alemtuzumab may effect MS via B-cell depletion. Neurol Neuroimmunol Neuroinflamm 2017;4:e360.

Neurology.org/nn © 2017 American Academy of Neurology 1