Volume 6, Number 2, April 2020 Neurology.org/NG

A peer-reviewed clinical and translational neurology open access journal

ARTICLE Heterozygous STUB1 missense variants cause ataxia, cognitive decline, and STUB1 mislocalization e397

ARTICLE ALS in Danish Registries: Heritability and links to psychiatric and cardiovascular disorders e398

ARTICLE

Heritability of cervical spinal cord structure e401

ARTICLE Use of local genetic ancestry to assess TOMM40-523’ and risk for Alzheimer disease e404 Academy Officers Neurology® is a registered trademark of the American Academy of Neurology (registration valid in the United States). James C. Stevens, MD, FAAN, President Neurology® Genetics (eISSN 2376-7839) is an open access journal published Orly Avitzur, MD, MBA, FAAN, President Elect online for the American Academy of Neurology, 201 Chicago Avenue, Ann H. Tilton, MD, FAAN, Vice President Minneapolis, MN 55415, by Wolters Kluwer Health, Inc. at 14700 Citicorp Drive, Bldg. 3, Hagerstown, MD 21742. Business offices are located at Two Carlayne E. Jackson, MD, FAAN, Secretary Commerce Square, 2001 Market Street, Philadelphia, PA 19103. Production offices are located at 351 West Camden Street, Baltimore, MD 21201-2436. Janis M. Miyasaki, MD, MEd, FRCPC, FAAN, Treasurer © 2020 American Academy of Neurology. Ralph L. Sacco, MD, MS, FAAN, Past President Neurology® Genetics is an official journal of the American Academy of Neurology. Journal website: Neurology.org/ng, AAN website: AAN.com CEO, American Academy of Neurology Copyright and Permission Information: Please go to the journal website (www.neurology.org/ng) and click the Permissions tab for the relevant Mary E. Post, MBA, CAE article. Alternatively, send an email to [email protected]. Chief Executive Officer General information about permissions can be found here: https://shop.lww.com/ journal-permission. 20l Chicago Ave Disclaimer: Opinions expressed by the authors and advertisers are not Minneapolis, MN 55415 necessarily those of the American Academy of Neurology, its affiliates, or of the Publisher. The American Academy of Neurology, its affiliates, and the Tel: 612-928-6000 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 Editorial Office or non-use of any of the material contained in this publication. Patricia K. Baskin, MS, Executive Editor Advertising Sales Representatives: Wolters Kluwer, 333 Seventh Avenue, Rachel A. Anderson, Administrative Assistant New York, NY 10001. Contacts: Eileen Henry, tel: 732-778-2261, fax: 973-215- 2485, [email protected] and in Europe: Craig Silver, tel: +44 Morgan S. Sorenson, Managing Editor 7855 062 550 or e-mail: [email protected]. fl Careers & Events: Monique McLaughlin, Wolters Kluwer, Two Commerce Neurology® Neuroimmunology & Neuroin ammation Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-521-8468, fax: 215- Neurology® Genetics 521-8801; [email protected]. Reprints: Meredith Edelman, Commercial Reprint Sales, Wolters Kluwer, Two Kathleen M. Pieper, Senior Managing Editor, Neurology® Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: 215-356-2721; Karen Skaja, Senior Editorial Coordinator [email protected]; [email protected]. Skyler M. Kane, Editorial Coordinator Special projects: US & Canada: Alan Moore, Wolters Kluwer, Two Commerce Square, 2001 Market Street, Philadelphia, PA 19103, tel: Margaret A. Rei, Editorial Coordinator 215-521-8638, [email protected]. International: Andrew Wible, Senior Manager, Rights, Licensing, and Partnerships, Wolters Kluwer; Lee Ann Kleffman, Managing Editor, Neurology® Clinical Practice [email protected]. Andrea Rahkola, Production Editor Kristen Swendsrud, Production Coordinator Sharon L. Quimby, Digital Managing Editor Kaitlyn Aman Ramm, Digital Multimedia/Graphics Coordinator Justin Daugherty, Digital Multimedia/Podcast Coordinator Madeleine Sendek, MPH, Digital Marketing Communications Coordinator

Publisher Wolters Kluwer Baltimore, MD

Publishing Staff Sarah Carrera, MBA, Executive Publisher Jessica Heise, Production Team Leader, Neurology Journals Megen Miller, Production Editor Steve Rose, Editorial Assistant Stacy Drossner, Production Associate

Copyright ª 2020 American Academy of Neurology. Unauthorized reproduction of this article is prohibited. A peer-reviewed clinical and translational neurology open access journal Neurology.org/NG

Neurology® Genetics

Editor Stefan M. Pulst, MD, Dr med, FAAN Vision Neurology®: Genetics will be the premier peer- reviewed journal in the field of neurogenetics. Deputy Editor Massimo Pandolfo, MD, FAAN Mission Neurology: Genetics will provide neurologists Associate Editors and clinical research scientists with Alexandra Durr, MD, PhD outstanding peer-reviewed articles, Suman Jayadev, MD editorials, and reviews to elucidate the role Margherita Milone, MD, PhD of genetic and epigenetic variations in Raymond P. Roos, MD, FAAN diseases and biological traits of the central Editorial Board and peripheral nervous systems. Hilary Coon, PhD Giovanni Coppola, MD ChantalDepondt, MD, PhD Editorial Tel: 612-928-6400 Brent L. Fogel, MD, PhD, FAAN Inquiries Toll-free: 800-957-3182 (US) AnthonyJ. Griswold, PhD Fax: 612-454-2748 Orhun H. Kantarci, MD [email protected] Julie R. Korenberg, PhD, MD Davide Pareyson, MD Shoji Tsuji, MD,PhD DinekeS. Verbeek,PhD Stay facebook.com/NeurologyGenetics Connected David Viskochil, MD,PhD twitter.com/greenjournal JulianeWinkelmann, MD Juan I. Young, PhD youtube.com/user/NeurologyJournal

instagram.com/aanbrain

Neurology® Journals

Editor-in-Chief Jos´eG.Merino,MD,MPhil,FAAN

Deputy Editor Classification of Evidence Olga Ciccarelli, MD, PhD, FRCP Review Team Melissa J. Armstrong, MD Section Editors Richard L. Barbano, MD,PhD, FAAN Biostatistics RichardM.Dubinsky,MD,MPH,FAAN Richard J. Kryscio, PhD Jeffrey J. Fletcher, MD, MSc Sue Leurgans, PhD Gary M. Franklin, MD, MPH, FAAN V. Shane Pankratz, PhD David S. Gloss II, MD,MPH&TM John J. Halperin, MD,FAAN fi Classi cation of Evidence Evaluations Jason Lazarou, MSc, MD Gary S. Gronseth, MD, FAAN Steven R. Mess´e, MD, FAAN Equity, Diversity, & Inclusion (EDI) Pushpa Narayanaswami, MBBS, DM, RoyH.Hamilton,MD,MS,FAAN FAAN Holly E. Hinson, MD, MCR, FAAN Alex Rae-Grant, MD

Podcasts Stacey L. Clardy, MD, PhD, FAAN Jeffrey B. Ratliff, MD, Deputy Podcast Editor

Ombudsman Jonathan W. Mink, MD, PhD, FAAN

Scientific Integrity Advisor David S. Knopman, MD, FAAN

Copyright ª 2020 American Academy of Neurology. Unauthorized reproduction of this article is prohibited. TABLE OF CONTENTS Volume 6, Number 2, April 2020 Neurology.org/NG

Articles e405 Genetic testing utilization for patients with neurologic disease and the limitations of claims data e397 Heterozygous STUB1 missense variants cause ataxia, S.J. Mackenzie, C.C. Lin, P.K. Todd, J.F. Burke, and B.C. Callaghan cognitive decline, and STUB1 mislocalization Open Access D.-H. Chen, C. Latimer, M. Yagi, M.K. Ndugga-Kabuye, E. Heigham, S. Jayadev, J.S. Meabon, C.M. Gomez, C.D. Keene, D.G. Cook, e406 SQSTM1 W.H. Raskind, and T.D. Bird Association of a structural variant within the Open Access gene with amyotrophic lateral sclerosis J. Pytte, R.S. Anderton, L.L. Flynn, F. Theunissen, L. Jiang, I. Pitout, e398 ALS in Danish Registries: Heritability and links to I. James, F.L. Mastaglia, A.M. Saunders, R. Bedlack, T. Siddique, N. Siddique, and P.A. Akkari psychiatric and cardiovascular disorders Open Access B.B. Trabjerg, F.C. Garton, W. van Rheenen, F. Fang, R.D. Henderson, P.B. Mortensen, E. Agerbo, and N.R. Wray e407 Novel EGR2 variant that associates with Open Access Charcot-Marie-Tooth disease when combined with α e399 MYORG-related disease is associated with central lipopolysaccharide-induced TNF- factor pontine calcifications and atypical parkinsonism T49M polymorphism ´ V. Chelban, M. Carecchio, G. Rea, A. Bowirrat, S. Kirmani, M.E. Blanco-Canto, N. Patel, S. Velasco-Aviles, A. Casillas-Bajo, ´ı ´ ´ı ´ı L. Magistrelli, S. Efthymiou, L. Schottlaender, J. Vandrovcova, J. Salas-Felipe, A. Garc a-Escriva, C. D az-Mar n, and H. Cabedo V. Salpietro, E. Salsano, D. Pareyson, L. Chiapparini, F. Jan, Open Access S. Ibrahim, F. Khan, Z. Qarnain, S. Groppa, N. Bajaj, B. Balint, K.P. Bhatia, A. Lees, P.J. Morrison, N.W. Wood, B. Garavaglia, and e408 Molecular diagnosis of muscular diseases in outpatient H. Houlden clinics: A Canadian perspective Open Access F. Thuriot, E. Gravel, C. Buote, M. Doyon, E. Lapointe, L. Marcoux, ´ ´ e401 Heritability of cervical spinal cord structure S. Larue, A. Nadeau, S. Chenier, P.J. Waters, P.-E. Jacques, S. Gravel, and S. L´evesque L. Solstrand Dahlberg, O. Viessmann, and C. Linnman Open Access e411 Hereditary cerebral amyloid angiopathy, Piedmont-type mutation e402 Mitochondrial diseases in North America: An analysis M.G. Kozberg, S.J. van Veluw, M.P. Frosch, and S.M. Greenberg of the NAMDC Registry Open Access E. Barca, Y. Long, V. Cooley, R. Schoenaker, V. Emmanuele, S. DiMauro, B.H. Cohen, A. Karaa, G.D. Vladutiu, R. Haas, J.L.K. Van Hove, F. Scaglia, e412 Clinical utility of multigene analysis in over 25,000 S. Parikh, J.K. Bedoyan, S.D. DeBrosse, R.H. Gavrilova, R.P. Saneto, G.M. Enns, P.W. Stacpoole, J. Ganesh, A. Larson, Z. Zolkipli-Cunningham, patients with neuromuscular disorders M.J. Falk, A.C. Goldstein, M. Tarnopolsky, A. Gropman, K. Camp, T.L. Winder, C.A. Tan, S. Klemm, H. White, J.M. Westbrook, J.Z. Wang, D.Krotoski,K.Engelstad,X.Q.Rosales,J.Kriger,J.Grier,R.Buchsbaum, A. Entezam, R. Truty, R.L. Nussbaum, E.M. McNally, and S. Aradhya J.L.P. Thompson, and M. Hirano Open Access Open Access e403 Characterization of the phenotype with cognitive Clinical/Scientific Notes impairment and protein mislocalization in SCA34 M. Beaudin, L. Sellami, C. Martel, L. Touzel-Deschˆenes, G. Houle, e400 Gerstmann-Str¨aussler-Scheinker disease (PRNP L. Martineau, K. Lacroix, A. Lavall´ee, N. Chrestian, G.A. Rouleau, F. Gros-Louis, R.J. Laforce, and N. Dupr´e p.D202N) presenting with atypical parkinsonism S. Baiardi, R. Rizzi, S. Capellari, A. Bartoletti-Stella, A. Zangrandi, Open Access F. Gasparini, E. Ghidoni, and P. Parchi e404 Use of local genetic ancestry to assess TOMM40-5239 Open Access and risk for Alzheimer disease e409 4H leukodystrophy: Mild clinical phenotype and P.L. Bussies, F. Rajabli, A. Griswold, D.A. Dorfsman, P. Whitehead, L.D. Adams, P.R. Mena, M. Cuccaro, J.L. Haines, G.S. Byrd, comorbidity with multiple sclerosis G.W. Beecham, M.A. Pericak-Vance, J.I. Young, and J.M. Vance S.M. DeGasperis, G. Bernard, N.I. Wolf, E. Miller, and D. Pohl Open Access Open Access

Copyright ª 2020 American Academy of Neurology. Unauthorized reproduction of this article is prohibited. TABLE OF CONTENTS Volume 6, Number 2, April 2020 Neurology.org/NG

P525L e410 Differential subcellular expression of FUS as a putative biomarker for ALS phenoconversion M. Caputo, V. La Bella, and A. Notaro Open Access e413 Multisystem mitochondrial disease caused by a rare m.10038G>A mitochondrial tRNAGly (MT-TG) variant O.V. Poole, A. Horga, S.A. Hardy, E. Bugiardini, C.E. Woodward, I.P. Hargreaves, A. Merve, R. Quinlivan, R.W. Taylor, M.G. Hanna, and R.D.S. Pitceathly Cover image Open Access From a high-resolution ex vivo MRI: 3D representation of cortical microbleeds and larger intracerebral hemorrhage location and insets with Correction scans of the cortical surface. See e411 e419 Heritability of cervical spinal cord structure

Copyright ª 2020 American Academy of Neurology. Unauthorized reproduction of this article is prohibited. ARTICLE OPEN ACCESS Heterozygous STUB1 missense variants cause ataxia, cognitive decline, and STUB1 mislocalization

Dong-Hui Chen, MD, PhD, Caitlin Latimer, MD, PhD, Mayumi Yagi, PhD, Mesaki Kenneth Ndugga-Kabuye, MD, Correspondence Elyana Heigham, BS, Suman Jayadev, MD, James S. Meabon, PhD, Christopher M. Gomez, MD, PhD, Dr. Bird [email protected] C. Dirk Keene, MD, PhD, David G. Cook, PhD, Wendy H. Raskind, MD, PhD, and Thomas D. Bird, MD

Neurol Genet 2020;6:e397. doi:10.1212/NXG.0000000000000397 Abstract Objective To identify the genetic cause of autosomal dominant ataxia complicated by behavioral ab- normalities, cognitive decline, and autism in 2 families and to characterize brain neuropatho- logic signatures of dominant STUB1-related ataxia and investigate the effects of pathogenic variants on STUB1 localization.

Methods Clinical and research-based exome sequencing was used to identify the causative variants for autosomal dominant ataxia in 2 families. Gross and microscopic neuropathologic evaluations were performed on the brains of 4 affected individuals in these families.

Results Mutations in STUB1 have been primarily associated with childhood-onset autosomal recessive ataxia, but here we report heterozygous missense variants in STUB1 (p.Ile53Thr and p.The37- Leu) confirming the recent reports of autosomal dominant inheritance. Cerebellar atrophy on imaging and cognitive deficits often preceded ataxia. Unique neuropathologic examination of the 4 brains showed the marked loss of Purkinje cells (PCs) without microscopic evidence of significant pathology outside the cerebellum. The normal pattern of polarized somatodendritic STUB1 protein expression in PCs was lost, resulting in aberrant STUB1 localization in the distal PC dendritic arbors.

Conclusions This study confirms a dominant inheritance pattern in STUB1-ataxia in addition to a recessive one and documents its association with cognitive and behavioral disability, including autism. In the most extensive analysis of cerebellar pathology in this disease, we demonstrate disruption of STUB1 protein in PCs as part of the underlying pathogenesis.

From the Department of Neurology (D.-H.C., E.H., S.J., T.D.B.), University of Washington, Seattle; Department of Pathology (C.L., C.D.K.), Neuropathology Division, University of Washington, Seattle; Geriatric Research, Education, and Clinical Center (GRECC) (M.Y., D.G.C., W.H.R., T.D.B.), VA Puget Sound Health Care System, Seattle, WA; Department of Medicine (M.K.N.-K., W.H.R., T.D.B.), Division of Medical Genetics, University of Washington, Seattle; Mental Illness Research, Education, and Clinical Center (MIRECC) (J.S.M., W.H.R.), VA Puget Sound Health Care System, Seattle, WA; Department of Psychiatry and Behavioral Sciences (J.S.M., W.H.R.), University of Washington, Seattle; Department of Neurology (C.M.G.), University of Chicago, IL; Department of Medicine (D.G.C.), Division of Gerontology and Geriatric Medicine, University of Washington, Seattle; and Department of Pharmacology (D.G.C.), University of Washington, Seattle.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary CADD = Combined Annotation Dependent Depletion; CHIP = C terminus of HSP 70 interacting protein; GERP = Genomic Evolutionary Rate Profiling; H&E = hematoxylin and eosin; PC = Purkinje cell; SCA = Spinocerebellar Ataxia; SCAR = Spinocerebellar Ataxia, Recessive; UW = University of Washington; WAIS-R = Wechsler Adult Intelligence Scale—Revised.

There are more than 40 genetic types of autosomal dominant Exome sequencing was performed at the UW Center for Pre- cerebellar ataxia referred to as the spinocerebellar ataxias cision Diagnostics on affected patients A III-1 and A IV-1. (SCAs).1 In addition, there are more than 100 autosomal re- Target enrichment by NimbleGen solution capture array and cessive disorders in which cerebellar ataxia is a prominent exome sequencing were performed and processed using Illu- feature, variably denoted as spinocerebellar ataxia, recessive mina HiSeq Analysis Software as previously described.8 The (SCAR) or ARCA (autosomal recessive cerebellar ataxia).1 reads were aligned and compared with the Homo Sapiens One recessive variety, SCAR-16, is caused by homozygous or GRCh37 reference genome. Sequences that failed Genome compound heterozygous mutations in the STUB1 (Stip1 ho- Analysis Toolkit quality filters were not further considered. mologous and Ubox-containing protein 1) gene.2,3 Most Filtering of the variants was based on (1) coding region se- individuals with SCAR-16 have an onset of a cerebellar syn- quence effect—missense, nonsense, coding indels, and splice drome in childhood or adolescence sometimes associated with sites; (2) heterozygosity, given the autosomal dominant in- cognitive disability. The single brain autopsy reported showed heritance pattern of the disease in the family; (3) shared by severe loss of Purkinje cells (PCs) and neurons of the granular both exomes or not read in one of the exomes (specified as “N” layer, accompanied by reactive Bergmann gliosis, but it lacked in the exome sequence data); (4) variant frequency—absence information about STUB1 staining in the cerebellum.4 Auto- from the 1000 Genomes Project database and frequency somal dominant inheritance of heterozygous STUB1 variants threshold less than 0.0001 in gnomAD; (5) high evolutionary have been recently been reported but without neuropathologic conservation, >3 by Genomic Evolutionary Rate Profiling evaluations.5,6 (GERP)9; (6) a predicted damaging or probably damaging effect on protein structure and function by Sorting Intolerant STUB1 protein, also called C terminus of HSP 70 interacting From Tolerant10 and PolyPhen11,12;and(7)CombinedAn- protein (CHIP), contains a tetratricopeptide repeat and notation Dependent Depletion (CADD) score13 >15. Variants a U-box. Among other functions, it is an E3 ubiquitin ligase/ that met these criteria were prioritized by gene expression, cochaperone that targets a broad range of chaperone protein function in cerebellum, information from animal models, and substrates, including Hsp70, Hsc70, Hsp90, and DNA poly- relevance to neurologic diseases. merase beta, for ubiquitin proteasome degradation.7 To confirm the exome variants and to investigate its cose- We report 2 families with autosomal dominant cerebellar ataxia, gregation with the disease in family A, we performed PCR progressive cognitive decline, and, in one individual, childhood capillary sequencing using customized primers to amplify the autism spectrum disorder associated with missense variants in fragment encompassing the candidate variants from genomic STUB1. The availability of 4 brain autopsies from these families DNA using Hot Start Taq Polymerase (Qiagen, Hilden, Ger- allowed a detailed evaluation of the cerebellar pathology and many) as previously described.14 revealed aberrant intracellular location of STUB1. Neuropathologic studies In both families, consent for autopsy was obtained from the Methods legal next of kin according to the protocols approved by the UW Institutional Review Board. At the time of autopsy, the Standard protocol approvals, registrations, brain was removed in the usual fashion. In family A, for patients and patient consents II-2 and III-1, the left halves were coronally sectioned and Recruitment and sampling of the families was approved by the frozen for possible biochemical studies and the right halves Institutional Review Board of the University of Washington were fixed in formalin. For patient A IV-1, the entire brain was (UW), and informed consent was provided by all patients. fixed in formalin. After fixation, the cerebrum was sectioned coronally, the brainstem was sectioned axially, and the cere- Genetics studies bellum sectioned sagittally. In family B, the brain was fixed in The pedigrees of the 2 families with familial cerebellar ataxia are 10% neutral buffered formalin. After fixation, the cerebrum was shown in figure 1, A and B. Clinical testing for the common sectioned coronally and the cerebellum and brainstem were hereditary ataxias was performed for the probands, III-1 in sectioned axially. family A and III-2 in family B, as described below in the clinical descriptions. Blood samples from the affected and unaffected A microtome was used to cut 4 μm-thick tissue sections from patients (indicated in the pedigree) were collected, and geno- formalin-fixed, paraffin-embedded tissue blocks. Hematoxylin mic DNA was extracted by standard methods. and eosin (H&E), Luxol fast blue (LFB), and Bielschowsky

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 1 Four generation pedigrees of 2 families with hereditary cerebellar ataxia and cognitive disability

If alive, current age in years is shown below the symbol. / = deceased; A = brain autopsy; ASD = autism spectrum disorder; black symbol = affected with ataxia; d = age at death. The variant or wild-type STUB1 allele is shown for individuals who were tested. silver stained slides were prepared. Using previously optimized performed in AR6 buffer for 15 minutes at 98°C. The sections conditions, immunohistochemistry was performed using were treated for 10 minutes in antibody diluent/block and a Leica Bond III Fully Automated IHC and ISH Staining incubated with primary antibody diluted in the same buffer. System (Leica Biosystems, Wetzlar, Germany). The sections The slides were washed in tris-buffered saline and polysorbate were immunostained with mouse monoclonal antibody against 20, incubated with the peroxidase-conjugated antimouse/rabbit paired helical filament tau (AT8, 1:1,000 dilution) (Pierce secondary antibody, and then washed and incubated with Opal Technology, Waltham, MA), mouse monoclonal against fluorescent substrate diluted in amplification buffer (Opal-650, β-amyloid (6E10, 1:5,000) (Covance, Princeton, NJ), rat Opal-520, and Opal-570, presented as pseudocolored immu- monoclonal against phosphorylated TDP-43 (ser409/ser410, nostaining in purple, green, and red, respectively). The sections 1:1,000) (Millipore, Burlington, MA), mouse monoclonal were washed and heat-treated to remove the antibody and then against α-synuclein (LB509, 1:500) (Invitrogen, Carlsbad, blocked and incubated with the second primary antibody as CA), rabbit polyclonal against glial fibrillar acidic protein followed by secondary and Opal substrate as before. The cycle (#Z033401-2, 1:2,000) (DAKO-Agilent, Santa Clara, CA), wasrepeatedoncemorewiththethirdprimaryantibody.After mouse monoclonal against calbindin protein (CB-955, 1: the third cycle, the sections were incubated brieflywithHoechst 1,000) (Sigma-Aldrich, St. Louis, MO), mouse monoclonal 33,256, rinsed with distilled H2O, and mounted with ProLong against ubiquitin (ubi-1, 1:50,000) (Millipore), and mouse Diamond Antifade Mountant (ThermoFisher, Waltham, MA). monoclonal against p62 (2C11, 1:4,000). Appropriate positive Confocal microscopy was performed using a Leica TCS SP5 and negative controls were included with each antibody and II microscope. Confocal images were acquired with the Leica each run. Application Suite and processed using identical data acqui- sition settings for the images in each specific figure. All STUB1 immunofluorescence microscopy confocal images are single z-plane scans. Postacquisition Paraffin-embedded brain sections were stained using reagents image processing and figure preparation were accomplished and protocols from a commercially available kit (Opal Manual using Leica Application Suite and Photoshop software that IHC Kit; Akoya Bioscience, Menlo Park, CA). The primary were limited to linear contrast and brightness adjustments antibodies used were: anti-calbindin D28k (1:1,000 dilution; only and were applied identically to all images regarding each EMD Millipore), anti-EAAT4 (1:200 dilution; Abcam, Cam- specific figure. bridge, United Kingdom), and anti-STUB1 (1:1,000 dilution; Abcam). Briefly, the sections were deparaffinized in xylene Data availability and rehydrated through graded alcohol and fixed in neutral- Deidentified data included in this study are available from the buffered formalin. Heat-mediated antigen retrieval was corresponding author on reasonable request.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Results difficulty with memory and marked problems with visual spatial skills, calculations, insight, and judgment. Wechsler Adult In- Clinical descriptions telligence Scale—Revised (WAIS-R) testing15 revealed a verbal Family A: The index case (III-1 in figure 1A) is a 60-year-old IQ of 74, performance IQ of 68, and full scale IQ of 71. He was man who developed cognitive problems and clumsiness in his labeled “demented.” He had a childlike affect with in- early 40s. He had 2 years of college education and was appropriate and tangential responses. MRI revealed marked employed as a mechanical engineer. Examination at 44 years of cerebellar atrophy. He experienced both alcohol and drug abuse age showed delayed but accurate responses to cognitive ques- and died at the age of 45 years without autopsy. tions, saccadic interruptions in smooth pursuit eye movements, and random brief jerking movements of his hands and legs. The daughter of the index patient (IV-1 in figure 1A) had an Repeat examination at 48 years of age showed a Mini-Mental unremarkable labor and delivery and reached developmental Status Examination score of 26 of 30, mild gait ataxia, moderate motor milestones of childhood within the range of normal. She dysarthria, irregular smooth pursuit and saccadic eye move- had average to below-average grades in school and was able to ments, and hyperactive tendon reflexes at the knees. A brain attend community college but dropped out of several courses. MRI revealed cerebellar atrophy that appeared out of pro- At 21 years of age, formal psychological testing showed a verbal portion to the relatively mild physical findings. Formal neu- IQ score of 84, performance score of 83, and full scale score of ropsychometric testing at 49 years of age revealed generally 83 (13th percentile; WAIS-R). Her highest score was in reading normal intelligence but abnormal scores for executive function recognition (30th percentile), but other subscale scores were and verbal memory (table 1). Repeat examination over the next below the 20th percentile. Special weaknesses were in digit 10 years showed slow and steady progression of his disease. His span, object assembly, arithmetic, and visual spatial analysis. She speech remained slurred, and his gait became more unsteady, had great difficulty with social interactions and was labeled with although he was still ambulatory with a walker. He had “extreme shyness” and avoidant personality disorder. She a stooped posture and was bradykinetic but had no rigidity and seemed “disconnected from reality,” was frequently teased, and no tremor. Repeat MRIs showed increased cerebellar atrophy had a “nervous breakdown” at 13 years of age. A detailed (figure 2A). He had cognitive decline, became less verbal, and analysis by a neuropsychologist at the age of 24 years demon- sometimes demonstrated agitation and belligerence. Psycho- strated that she fell within the autism spectrum disorder. WAIS- metric testing documented deterioration with deficiencies in R IQ and memory testing was similar to the age of 21 years, verbal and visual memory, general language ability, processing with mental arithmetic, comprehension, block design, object speed, executive functions, and abstract reasoning with many assembly, and visual memory at or below the fifth percentile subtests in the 1st to 10th percentile. Medical comorbidities (table 1). Neurologic examination at the age of 24 years was included adult-onset diabetes mellitus and central sleep apnea remarkable only for occasional saccadic interruptions in smooth treated with nocturnal continuous positive airway pressure. pursuit and mild deep tendon hyper-reflexia. She had no nys- Medications included olanzapine, lorazepam, Neurontin, and tagmus, no dysarthria, no intention tremor, and no gait ataxia. baclofen. Molecular testing for SCAs 1, 2, 3, 6, 7, 8, 14, and 12, However, because of her family history of hereditary ataxia, she and DRPLA were negative. He died at the age of 60 years in an received a brain MRI that showed definite cerebellar atrophy. assisted living facility, and his brain autopsy was obtained. Over the next 10 years, there was no change in her neurologic examination and she did not develop ataxia. A repeat brain MRI The father (II-2 in figure 1A) of the index patient had an onset at 32 years of age showed mild increase in the degree of cere- of cognitive problems and poor coordination in his early 60s. bellar atrophy and mild generalized cerebral volume loss that He had been a pilot, mechanic, and homebuilder. A single was greater than expected for her age (figure 2B). She moved neurologic examination at 68 years of age revealed mild gait out of state and began to deteriorate in her late 30s and early ataxia, dysarthria, poor hand coordination, abnormal saccadic 40s. She developed gait ataxia and started using a wheelchair. eye movements, hyperactive knee reflexes, and extensor plantar She developed severe cognitive decline with confusion, disori- reflexes. He had intact short-term memory and could do simple entation, poor judgment, and deficient problem solving. She arithmetic but exhibited paraphasias, inability to perform developed bladder incontinence and became fully dependent a three-step command or do a word problem in his head, and on nursing care. She died at the age of 43 years, and her brain had concrete thinking and poor performance of the Luria hand autopsy was obtained. Her brother (IV-2) had a normal ex- sequence. His ataxia and cognitive problems progressed, and he amination and normal brain MRI at the age of 40 years. started using a wheelchair. He died at the age of 79 years, and a brain autopsy was obtained. Family B: The index patient (III-2 in figure 1B) was a woman who developed difficulty with coordination, mental confusion, The brother (III-3 in figure 1A) of the index patient had a ju- and a personality change at approximately 40 years of age. nior college education. He had an onset of poor coordination in Neurologic examination at the age of 42 years revealed gait his late 20s. Examination at 34 years of age revealed marked gait ataxia, dysarthria, gaze-evoked horizontal nystagmus, and dys- ataxia, saccadic smooth pursuit eye movements with hyper- metria. Tendon reflexes and sensory examination were un- metric saccades, dysarthria, dysmetria, and hyperactive tendon remarkable. Brain MRI at the age of 48 years showed moderate reflexes. Detailed mental status examination revealed moderate cerebellar atrophy (figure 2, C and D). The Mini-Mental Status

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 1 Neuropsychological testing Table 1 Neuropsychological testing (continued)

Family A IV-1, age 24 Family B III-1, age 42

Wechsler Adult Wechsler Adult Controlled oral Intelligence Wechsler Memory Intelligence Scale word association Scale—Revised15 Scale Revised43 Revised15 test49

Verbal IQ 78 (7%) Verbal memory 27% Wechsler memory Wisconsin card index scale sorting test48

Performance IQ 88 Visual memory 4% Verbal immediate 95% Categories 16% (21%) index Verbal delayed 75% No. presev errors 30% Full scale IQ 80 (9%) General memory 14% index Visual immediate 37% No. error 34%

Mental arithmetic 5% Attention/ 2% Visual delayed 25% concentration index Hooper visual 20% Comprehension 5% Delayed memory 30% California verbal learning organization index test50 test49

Vocabulary 25% Immediate recall 29% Trial 1 2%

Block design 5% Delayed recall 22% Trial 5 <1%

Object assembly 2% Digits forward 12% B List 2%

Digit symbols 50% Digits backward 26% Immediate recall 14%

Digit span 38% Visual memory span 10% Cued 14% forward Delayed recall 2% Similarities 9% Trail-making test A44 80% Cued 14% Information 50% Trail-making test B 10% Recognition 14% Picture completion 84% Intrusions 85% Picture arrangement 75%

Family A III-1, age 49 Examination score was 28 of 30. Detailed formal neuro- Wide-range assessment of psychological testing revealed intact verbal memory but with Woodcock Johnson tests memory and fi 45,46 47 numerous cognitive test scores at or below the fth percentile, of cognitive ability learning and she was noted to have marked general difficulties in the Board cognitive ability 12% Picture memory 63% management and production of cognitive processes (table 1).

Memory for name 43% Design memory 9% She was a college graduate and had been a schoolteacher but was no longer able to perform her teaching duties. Multiple Memory for sentences 1% Verbal memory 9% genetic tests for ataxia were all normal including SCAs 1, 2, 3, 5, Visual matching 2% Story memory 2% 6, 7, 10, 14, 21, dentatorubral-pallidoluysian atrophy, Frie- dreich’s ataxia, and senataxin. Incomplete words 20% Wisconsin card sorting48 Her family history was positive for additional affected people. Visual closure 44% All subtests 2%– 12% Her father had died at the age of 75 years with ataxia, and her paternal aunt died at the age of 73 years and had been using Picture vocabulary 72% a wheelchair for 10 years with ataxia. Her paternal grandmother Analysis synthesis 33% died at the age of 68 years after admission to a state mental

Family B III-1, age 42 institution with ataxia, cognitive decline, and psychotic features. The patient had younger siblings who were not symptomatic, Wechsler Adult Controlled oral Intelligence Scale word association one of whom (III-3) had a normal examination at the age of 56 Revised15 test49 years and remains unaffected at the age of 67 years.

Digit span <10% FAS <1% The patient had a slowly progressive course, eventually re- Picture completion 37% Animals <1% quiring a wheelchair. Her personality change included episodes Picture arrangement 50% of agitation and belligerence treated with olanzapine. Later in life, she developed a large excursion movement disorder of her Block design 10% upper and lower limbs that was described as an unusual wing-

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Figure 2 Neuroimaging

(A) MRI of patient III-1 in family A at 53 years of age showing marked midline cerebellar atrophy, mildly enlarged lateral ventricle, and normal corpus callosum. (B) MRI of patient A IV-1 at 32 years of age showing marked midline cerebellar atrophy before the onset of ataxia but having cognitive disability and autism spectrum disor- der. (C and D) MRI of patient III-2 in family B at 48 years of age showing midline (C) and lateral (D) cerebellar atrophy.

beating tremor. She died at the age of 61 years, and her brain variant is not present in the gnomAD database. STUB1 has been autopsy was obtained. associated with SCAR16, an autosomal recessive SCA.2,17 Sanger sequencing confirmed the variant and identified it in all 4 affected Genetic studies patients and not in the unaffected one. None of the other 3 Whole exome sequence data were obtained from the individuals prioritized candidate variants showed complete cosegregation III-1 and IV-1 in family A. The average allele depth of coverage with the disease in family A. In family B, clinical exome se- was 100× with more than >99% of the exome covered at more quencing reported a heterozygous STUB1 variant of uncertain than 20×. We first scanned the exome data for variants in genes significance c.111C>G, p.The37Leu in the proband. The variant for autosomal dominant SCAs; no potential pathogenic variants is located in a highly conserved residue (GERP 3.5; phastCons were present. Both exomes contained a small CTG repeat ex- 1) and is predicted to be disease causing by SIFT (0.01), Poly- pansion (9CTG/13CTG) in ATXN3, but the same allele was Phen2 (1.00), and CADD (24.7). Furthermore, the variant is not carried by the unaffected patient (III-2) and is reported as present in the gnomAD database. This variant was not detected a benign polymorphism in ClinVar. We then followed a stepwise by targeted sequencing in the unaffected sibling. Other members filtering protocol to prioritize candidate causative variants. Based of these 2 families were not available for testing. on the autosomal dominant inheritance pattern of the disease in the family and the population prevalence of 1–5/100,000 for We performed a literature search using PubMed and multiple combined SCAs,1,16 we selected heterozygous variants detected public variant databases including ClinVar, MARRVEL,18 in both exomes with a conservative minor allele frequency Varsome,19 and VarCards20 for variants reported in STUB1 threshold of <0.0001 to avoid overfiltering. Other criteria in- (figure 3). Considering the phenotypes, evidence for cose- cluded high/moderate functional effect based on variant type gregation, and predicted or experimentally shown effect on (missense, nonsense, coding indels, and splice site) and pre- protein function, we found 26 homozygous or compound diction algorithms and high evolutionary conservation. heterozygous-likely pathogenic variants, all of which were reported in SCAR16 cases; none of these STUB1 variants were This process left 4 candidate variants (table e-1, links.lww.com/ reported in autosomal dominant SCA. Most of these mutations NXG/A222), of which the missense variant in STUB1 was the are included in a recent report on SCAR16.21 While we were most compelling. The variant in exon 1, c.158T>C, p.Ile53Thr preparing this manuscript, one case with SCA and a frame-shift (RefSeq NM_005861) affects an evolutionarily conserved resi- mutation p.L275Dfs*16 and 2 with missense mutations p.G33S due (GERP 3.66, phastCons 1) and is predicted to be deleterious and p.P228S in STUB1 were reported.5,6 This deletion and by SIFT (0.001), PolyPhen2 (1.00), and CADD (24.1). The p.P228S are located in the U-box domain, whereas the 2

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 3 Ataxia-related variants in STUB1

(A) STUB1 gene showing reported variants associated with autosomal dominant and recessive ataxias and the corresponding locations in the protein. Colors indicate the TPR (turquoise), coiled-coil (purple), and E3 ubiquitin ligase (U-box) domains in STUB1. Bold = heterozygous autosomal dominant variants reported by Genis et al5 and De Michele et al6 and in this report (in red). Arrow = an intronic mutation at the donor splice site of exon 4. (B) Diagram depicts the location of the p.Ile53Thr (red amino acid) and p.Phe37Leu (yellow amino acid) missense variants within a 3-D model of a STUB1 homodimer (the functional isoform). STUB1 was modeled with PyMOL (pymol.org) using the data from the protein data bank (rcsb.org, PDB ID#2C2L). TPR = tetratricopeptide repeat. variants in families A and B are both in the N-terminal tetra- A221). The cerebellar atrophy globally included the cortex tricopeptide repeat domain (also p.G33S, figure 3). with marked thinning of the folia but the deep cerebellar nuclei appeared unremarkable (figure e-1). Neuropathologic studies Patient B III-2 Gross findings The 1,040 g (fresh) brain had no significant cortical atrophy; ff Patient A II-2 however, the cerebellum was markedly atrophic in a di use global pattern. In addition, the basis pontis was mildly The brain only showed atrophy of the superior vermis and fl dorsal region of the cerebellar hemisphere. No other signifi- attened. cant pathologic findings were noted. Histopathologic findings Patient A III-1 Each case in family A showed similar histologic findings in The 1,120 g (fresh) brain had mild cortical atrophy involving the cerebellum (figure 4). H&E/LFB-stained slides show frontal lobes without significant temporal, parietal, or occipital thinning of the cerebellar folia with marked loss of PCs. cortical atrophy. On cut sections, the lateral ventricles were Bielschowsky silver stain demonstrates empty baskets in enlarged bilaterally and the hippocampus was moderately association with the loss of PCs, and there is reduced cal- atrophic. The cerebellar vermis was small and firm, and the bindin immunohistochemistry. There is also increased dentate nucleus appeared mildly atrophic on cut sections. astrogliosis and Bergmann gliosis highlighted on the GFAP immunostain. Scattered ubiquitinated inclusions are pres- Patient A IV-1 ent, but no p62 inclusions are identified. Notably, there The 939 g (fresh) brain had moderate cortical atrophy and was no evidence of TDP-43 or alpha-synuclein-positive severe cerebellar atrophy (figure e-1, links.lww.com/NXG/ inclusions.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 Figure 4 Microscopic neuropathology

Microscopic sections of the patients from both families showing marked cerebellar atrophy with thinning of the folia (H&E/LFB low magnification) and Purkinje cell (PC) loss (H&E/LFB high magnification). The loss of PCs is also highlighted by the empty baskets noted on Bielschowsky silver stain and bythe severe reduction in calbindin-positive fibers (calbindin and GFAP immunostaining available on request). H&E = hematoxylin and eosin.

In patient A II-2, other neuropathologic findings including and astrogliosis. Scattered ubiquitin-positive inclusions are neurofibrillary tangles present within several regions of the noted, but no p62- TDP-43-positive or alpha-synuclein- hippocampus, including CA2 and subiculum, in addition to the positive inclusions are identified. Other neuropathologic transentorhinal cortex (Braak III) without evidence of beta- changes include sparse neuritic plaques in the neocortex amyloid pathology (Thal 0, Consortium to Establish a Registry (CERAD sparse) and diffuse amyloid plaques present in for Alzheimer’s Disease [CERDAD] score absent). In patient A isocortex, hippocampus, and striatum (Thal 3). Neurofi- III-1, scattered neurofibrillary tangles were limited to the pre- brillary tangles are limited to the pre-alpha cells of the alpha cells of the transentorhinal cortex (Braak I) with patchy, transentorhinal cortex (Braak I). No pathologic changes diffuse neocortical amyloid plaques (Thal 1), and no neuritic were noted in the basal ganglia. plaques were identified (CERAD absent). In patient A IV-1, there are rare neurofibrillary tangles limited to the pre-alpha cells STUB1 confocal microscopy findings of the transentorhinal cortex without the evidence of beta- There are at least 29 previously reported pathogenic variants in amyloid pathology (Braak I, Thal 0, CERAD absent). There STUB1 (figure 3), 3 dominant and 26 recessive, some of which were also patchy white matter lesions located in the subcortical destabilize the protein in vitro,3 but currently, there are no white matter, corpus callosum, anterior commissure, and the reported data regarding the effects of mutant STUB1 protein internal capsule characterized by a loss of myelinated axons, expression in the cerebellum of an individual harboring an most consistent with acute microinfarcts. No pathologic findings ataxia-related STUB1 variant. To address this question, we were noted in the basal ganglia in any of the cases in this family. carried out confocal microscopy using the rare neuropathologic cerebellum specimens from our 4 autopsied cases (II-2, III-1, In patient B III-2, H&E-stained slides show diffuse, bilateral and IV-1 in family A and III-2 in family B). PC loss and cerebellar cortical atrophy. Bielschowsky stains confirm PC loss highlighted by diffuse empty baskets and Figure 5A shows results from 2 normal controls (a–h), where rare torpedo axons. Calbindin immunostain showed marked the STUB1 expression in the cerebellar molecular layer was loss of fibers, and GFAP immunostain highlighted Bergman expressed prominently in PCs. Additional information about

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 5 Loss of polarized STUB1 expression in Purkinje cells in patients with STUB1-cerebellar ataxia

Immunostaining was performed with antibodies to STUB1 (pseudocolored purple), EAAT4 (green), and calbindin (red), and all scanning acquisition and postprocessing parameters were car- ried out identically among the cases about the controls in each panel. In 3 different normal control individuals (A.a–A.d, B.a–B.d, C.a–C.d), STUB1 im- munoreactivity was uniformly localized primarily in PC cell bodies and proximal dendrites, with little expression in distal PC dendritic arbors. (A and B) Family A with STUB1-Ile53Thr. Loss of STUB1 polarization results in aberrant ex- pression in distal PC dendritic arbors and cell bodies in patients III-1 (A.e–A.h), IV-1 (B.e–B.h), and II-2 (B.i–B.l). Scale bars = 100 μm. (C) Family B with STUB1- Phe37Leu. In individual III-2, STUB1 was aberrantly expressed in PC distal arbors (C.e–C.h). Scale bars = 50 μm.

the localization of STUB1 was revealed by coimmunostaining out to augment the calbindin staining because EAAT4 more calbindin and the neuronal glutamate transporter EAAT4, comprehensively labels distal dendritic compartments of the which is known to be expressed virtually exclusively in PC PCs (compare b and c in figure 5A). In a normal control, plasma membranes.22,23 EAAT4 immunostaining was carried STUB1 expression was polarized in PCs, with the most

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 prominent expression confined to cell bodies and proximal processes, we also examined STUB1 expression in SCA5, dendrites (somatodendritic) and much less in the distal den- SCA3, and SCA-7. In SCA5 (figure 6, E–H) and SCA3 dritic compartments that project prominently into the cere- (figure 6, I–L), STUB1 immunostaining was not different bellar molecular layer. Note that although EAAT4 from the control (figure 6, A–D). These findings suggest that immunostaining (figure 5A.b) confirmed that the image field aberrant STUB1 expression in PCs is not mediated solely by included extensive normal-appearing PC dendritic arbors, nonspecific PC degeneration. However, in the SCA7 case STUB1 expression was restricted mostly to the PC cell bodies (figure 6, M–P), STUB1 appeared mislocalized in distal and proximal dendrites (figure 5A.a). In marked contrast to dendritic arbors. Thus, it is unlikely that PC STUB1 mis- this, STUB1 immunostaining from patient III-1 in family A localization is mediated selectively by these 2 dominantly with Ile53Thr-STUB1 (figure 5A.i–l) revealed an apparent loss inherited STUB1 variants. of normal STUB1 polarization that was evidenced by expres- sion throughout the distal PC arbors in addition to the soma- todendritic compartments. As with patient A III-1, STUB1 Discussion immunoreactivity was aberrantly expressed in distal PC den- dritic processes of patient A IV-1 (figure 5B.e–h) and A II-2 STUB1 was previously associated with SCAR16, an autosomal (figure 5B.i–l). Inspection of the cerebellum from family B recessive ataxia.3 Recently, 3 families with ataxia segregating patient III-2 with Phe37Leu-STUB1 (figure 5C.e–h) also heterozygous frameshift (p.L275Dfs*16) or a missense revealed the same pattern of aberrant distal PC arbor STUB1 (p.Gly33Ser or p.Pro228Ser) variant in STUB1 have been expression. reported, but neuropathology was not included.5,6 The 2 new families we describe herein confirm that heterozygous variants To address the possibility that loss of polarized PC STUB1 in STUB1 can cause autosomal dominant hereditary cere- expression could be due to nonspecific neurodegenerative bellar ataxia, expand the associated pathogenic missense

Figure 6 STUB1 expression pattern in Purkinje cells in a number of non-STUB1-associated ataxias and cerebellar degeneration

In a normal individual (A–D), STUB1 immunoreactivity was expressed mostly in PC cell bodies and proximal dendrites (pseudocolored purple). Calbindin (red) and EAAT4 (green) immunostained PC bodies and den- drites. STUB1 localization appeared normal in PCs from SCA5 (E–H) and SCA3 (I–L). In an SCA7 case, STUB1 expression was aberrantly localized in distal PC dendritic arbors (M–P). Scale bars = 100 μm.

10 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG variants, broaden the age at onset range, and neuro- This is consistent with the hypothesis that the cerebellum developmental manifestations to include childhood autism participates in cognitive processing and emotional spectrum disorder, document that the cerebellum is the control.37,38 The only previous neuropathologic description prominent target of the pathologic process, and provide evi- associated with SCA-STUB1 disease is in an autosomal re- dence that there is a change in cellular distribution of STUB1 cessive case that also showed marked PC loss.4 CHIP in PCs. Taken together, these 5 families document that (STUB1) immunostaining labeled primarily cytoplasm in STUB1-related dominant ataxia is commonly associated with the neurons of the paraventricular nuclei of the hypothala- cognitive and behavioral disturbances that may precede the mus and nuclear staining of pyramidal neurons of the medial- ataxia by years. entorhinal cortex. No CHIP staining of the cerebellum was reported. The authors speculated that the involvement of With advances in sequencing and improvements in variant frontal and temporal cortex was likely related to the dementia interpretation techniques, the tremendous heterogeneity of that developed in the later stages of the disease. The remarkably neurogenetic disorders is more evident and it is increasingly low total brain weight in our patient A IV-1 suggests there may apparent that genes once associated only with one form of be abnormalities in the cerebral cortex beyond what we were inheritance are in fact capable of causing both recessive and able to detect microscopically and may be related to the – dominant forms.24 27 The nomenclature that assigns a se- hypometabolism on fluorodeoxyglucose-PET studies of the quential number to each gene in a disease category provides cerebellum, striatum, and cerebral cortex reported.6 no clarifying information and has limited clinical or research utility. As proposed, regarding the nomenclature of re- The CHIP ubiquitin ligase encoded by STUB1 functions as cessive cerebellar ataxia28 and other genetic movement a homodimer.39 A number of variants responsible for the disorders,29 we suggest ATX-STUB1 and SCA-STUB1 to recessive form of ATX-STUB1 have been shown to reduce replace SCAR16 and SCA48. STUB1 protein stability in vitro, suggesting that these mutations may lead to lower steady-state STUB1 expression The features of the recessive and dominant forms of SCA- levels,3 yet the carriers remain unaffected. Therefore, hap- STUB1 ataxia overlap, including progressive cognitive loinsufficiency and dominant negative mechanisms seem impairment. However, in the recessive form, the behavioral unlikely for the dominant presentation. The autosomal deficit is usually mild and the onset of ataxia is usually in dominant missense and frameshift variants may have a toxic early childhood to the mid-20s.2,30,31 Additional features of gain of function effect.5,6 the recessive form may include spasticity of the lower limbs, mild peripheral sensory neuropathy, and hypogonado- In support of a toxic gain of function, we found that the tropic hypogonadism.30,32,33 As with other dominant dis- intracellular location of STUB1 protein in PCs is altered by orders, clinical manifestations and age at onset are variable the missense variants. In PCs, STUB1 is normally polar- even within families. The mild parkinsonian features noted ized, with the most prominent expression in somatoden- in III-1 (family A) and the wing beating-like tremor seen dritic compartments and lower expression in the distal late in the disease in III-2 (family B) are apparently addi- dendritic arbors that extend into the cerebellar molecular tional features that may occur in the dominant form of this layer. The molecular layer is filled with a dense network of syndrome. EAAT4-positive “knotted lace-like” PC arbors that is quite large compared with the PC somatodendritic compart- A role for the cerebellum in autism has been previously sug- ments, hence further emphasizing the highly polarized gested, and both decreased cerebellar vermis volume and PC nature of normal STUB1 protein expression in PCs. In the – density have been observed.34 36 The single case of autism p.Ile53Thr and p.Phe37Leu cases in this study, STUB1 spectrum disorder in the present family does not prove an appeared to lose somatodendritic polarization, allowing association with STUB1 mutations but is noteworthy and STUB1 to localize in distal PC arbors and cell bodies. The continued search for additional examples in other families. It is faint cytoplasmic staining of STUB1 and extension into of interest that this family member’s MRI showed cerebellar dendrites and axons were also observed in the cortex of the atrophy at the time of her diagnosis of autism but before the patient with recessive ATX-STUB14; however, the cere- onset of ataxia and cognitive decline. bellum was not evaluated for STUB1 in that study. Addi- tional functional studies of the pathogenic variants The neuropathologic findings in the 4 brains from the 2 responsible for these ataxias may clarify the biochemical families reported here are remarkably similar in the prom- consequences of this location shift. inent cerebellar atrophy with the loss of PCs. Our cases support the direct effect of cerebellar pathology on the The involvement of STUB1 in disease extends beyond its cognitive/behavioral features of SCA-STUB1 because cere- causative role in STUB1-ataxia. The CHIP and ataxin-1 pro- bellar atrophy was found on imaging when the cognitive teins directly interact, and CHIP promotes ubiquitination of behavioral abnormality was evident before the clinical signs expanded ataxin-1.40 Ataxin-3 is involved in the degradation of of ataxia were displayed and because no significant brain misfolded chaperone substrates through its interaction with pathology outside the cerebellum was identified on autopsy. STUB1/CHIP.41 Ataxin-3 is recruited to monoubiquitinated

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 11 STUB1/CHIP where it prevents further ubiquitin chain ex- tension. These reports led us to examine STUB1 mislocaliza- Appendix Authors

tion in other type of SCAs. The shift in subcellular localization Name Location Role Contribution in our patients was not seen in SCA3, SCA5, or a nongenetic case with chronic ischemic cerebellar injury (data not shown). Dong-Hui University of Author Designed and Chen, MD, Washington, Seattle conceptualized the Taken together, these findings argue against the possibility that PhD study, collected and STUB1 mislocalization is mediated nonspecifically by degen- analyzed the data, drafted the erating PCs. However, we also found evidence of disturbed manuscript for STUB1 polarization in the SCA7 case. Therefore, it is unlikely intellectual content, and edited the that STUB1 mislocalization is restricted to these dominantly manuscript expressed STUB1 mutations. Detailed study of more cases and Caitlin University of Author Collected and more subtypes of SCA may shed light on the possible role Latimer, Washington, Seattle analyzed the data and of STUB1 in a common pathway with relevance to SCA MD, PhD drafted the manuscript for development. intellectual content

Mayumi VA Puget Sound Author Collected and It is also noteworthy that the 4 brains from the 2 families we Yagi, PhD Health Care System, analyzed the data report demonstrated no tau pathology other than low Braak Seattle ’ stages consistent with the patients ages.Thisisincontrast Mesaki University of Author Collected and to a report of substantial tau pathology in CHIP/STUB1 Kenneth Washington, Seattle analyzed the data and knockout mice and Caenorhabditis elegans consistent with Ndugga- edited the manuscript Kabuye, MD this protein being involved in the ubiquitination of tau.42 This contrast is additional evidence that the missense Elyana University of Author Collected and Heigham Washington, Seattle analyzed the data mutations reported here are unlikely to cause a loss of function. Suman University of Author Collected the data Jayadev, MD Washington, Seattle

In conclusion, a variety of mutations in STUB1 can cause either James S. VA Puget Sound Author Analyzed the data Meabon, Health Care System, autosomal recessive or autosomal dominant cerebellar ataxia PhD Seattle often associated with a prominent cognitive affective syndrome Christopher University of Chicago Author Collected the data that may include the autism spectrum disorder. The dominant M. Gomez, form presents with cognitive impairment at an earlier stage and MD, PhD ataxia at a later stage, in comparison to the recessive form. The C. Dirk University of Author Collected and brunt of the pathologic process affects PCs where there is Keene, MD, Washington, Seattle analyzed the data a mislocalization of the STUB1 protein from the cytoplasm to PhD the dendritic arbor. Further resolution of this pathologic pro- David G. VA Puget Sound Author Conceptualized the cess will help explain the extensive variability of the hereditary Cook, PhD Health Care System, study; collected and Seattle analyzed the data ataxias. Wendy H. University of Author Designed and Acknowledgment Raskind, Washington, Seattle conceptualized the MD, PhD study, drafted the The authors are grateful to the families whose participation manuscript for made this work possible and to John Wolff, Allison Beller, Lisa intellectual content, and edited the Keene, Kim Howard, and Emily Trittschuh for their excellent manuscript technical assistance. Thomas D. University of Author Conceptualized the Bird, MD Washington, VA study, drafted the Study funding Puget Sound Health manuscript for Care System, Seattle intellectual content, Supported by NIH (1R01NS069719, U01AG005136, and and edited the 5T32GM007454) and the United States (U.S.) Department of manuscript Veterans Affairs (Merit Review Award Number 101 CX001702 from the Clinical Sciences R&D [CSRD] Service and RDIS# 0005 from the Office of Research and Development Medical References 1. Bird T. Hereditary ataxia overview. In: GeneReviews at GeneTests: Medical Research Service). Genetics Information Resource [Database Online] 1993-2019. Seattle: University of Washington; 2019. 2. Depondt C, Donatello S, Simonis N, et al. Autosomal recessive cerebel- Disclosure lar ataxia of adult onset due to STUB1 mutations. Neurology 2014;82: Disclosures available: Neurology.org/NG. 1749–1750. 3. Kanack AJ, Newsom OJ, Scaglione KM. Most mutations that cause spinocerebellar ataxia autosomal recessive type 16 (SCAR16) destabilize the protein quality-control Publication history E3 ligase CHIP. J Biol Chem 2018;293:2735–2743. fi 4. Bettencourt C, de Yebenes JG, Lopez-Sendon JL, et al. Clinical and neuropathological Received by Neurology: Genetics October 16, 2019. Accepted in nal features of spastic ataxia in a Spanish family with novel compound heterozygous form November 7, 2019. mutations in STUB1. Cerebellum 2015;14:378–381.

12 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 5. Genis D, Ortega-Cubero S, San Nicolas H, et al. Heterozygous STUB1 mutation causes 27. Morimoto Y, Yoshida S, Kinoshita A, et al. Nonsense mutation in CFAP43 causes normal- familial ataxia with cognitive affective syndrome (SCA48). Neurology 2018;91:e1988–e1998. pressure hydrocephalus with ciliary abnormalities. Neurology 2019;92:e2364–e2374. 6. De Michele G, Lieto M, Galatolo D, et al. Spinocerebellar ataxia 48 presenting with 28. Rossi M, Anheim M, Durr A, et al. The genetic nomenclature of recessive cerebellar ataxia associated with cognitive, psychiatric, and extrapyramidal features: a report of ataxias. Mov Disord 2018;33:1056–1076. two Italian families. Parkinsonism Relat Disord 2019;65:91–96. 29. Marras C, Lang A, van de Warrenburg BP, et al. Nomenclature of genetic movement 7. McDonough H, Patterson C. CHIP: a link between the chaperone and proteasome disorders: recommendations of the international Parkinson and movement disorder systems. Cell Stress Chaperones 2003;8:303–308. society task force. Mov Disord 2016;31:436–457. 8. Chen D-H, Below JE, Shimamura A, et al. Ataxia-pancytopenia syndrome is caused by 30. Shi Y, Wang J, Li JD, et al. Identification of CHIP as a novel causative gene for missense mutations in SAMD9L. Am J Hum Genet 2016;98:1146–1158. autosomal recessive cerebellar ataxia. PLoS One 2013;8:e81884. 9. Davydov EV, Goode DL, Sirota M, Cooper GM, Sidow A, Batzoglou S. Identifying 31. Heimdal K, Sanchez-Guixe M, Aukrust I, et al. STUB1 mutations in autosomal a high fraction of the human genome to be under selective constraint using GERP++. recessive ataxias—evidence for mutation-specific clinical heterogeneity. Orphanet J PLoS Comput Biol 2010;6:e1001025. Rare Dis 2014;9:146. 10. Ng PC, Henikoff S. SIFT: predicting amino acid changes that affect protein function. 32. Shi CH, Schisler JC, Rubel CE, et al. Ataxia and hypogonadism caused by the loss of Nucleic Acids Res 2003;31:3812–3814. ubiquitin ligase activity of the U box protein CHIP. Hum Mol Genet 2014;23: 11. Sunyaev S, Ramensky V, Koch I, Lathe W III, Kondrashov AS, Bork P. Prediction of 1013–1024. deleterious human alleles. Hum Mol Genet 2001;10:591–597. 33. Synofzik M, Schule R, Schulze M, et al. Phenotype and frequency of STUB1 muta- 12. Ramensky V, Bork P, Sunyaev S. Human non-synonymous SNPs: server and survey. tions: next-generation screenings in Caucasian ataxia and spastic paraplegia cohorts. Nucleic Acids Res 2002;30:3894–3900. Orphanet J Rare Dis 2014;9:57. 13. Kircher M, Witten DM, Jain P, O’Roak BJ, Cooper GM, Shendure J. A general 34. Webb SJ, Sparks BF, Friedman SD, et al. Cerebellar vermal volumes and behavioral framework for estimating the relative pathogenicity of human genetic variants. Nat correlates in children with autism spectrum disorder. Psychiatry Res 2009;172:61–67. Genet 2014;46:310–315. 35. Skefos J, Cummings C, Enzer K, et al. Regional alterations in purkinje cell density in 14. Chen YZ, Matsushita MM, Robertson P, et al. Autosomal dominant familial dyski- patients with autism. PLoS One 2014;9:e81255. nesia and facial myokymia: single exome sequencing identifies a mutation in adenylyl 36. Wang SS, Kloth AD, Badura A. The cerebellum, sensitive periods, and autism. Neuron cyclase 5. Arch Neurol 2012;69:630–635. 2014;83:518–532. 15. Wechsler D. Wechsler Adult Intelligence Scale—Revised (WAIS-R). San Antonio: 37. Schmahmann JD. The cerebellum and cognition. Neurosci Lett 2019;688:62–75. The Psychological Corporation; 1981. 38. Liang KJ, Carlson ES. Resistance, vulnerability and resilience: a review of the cognitive 16. Ruano L, Melo C, Silva MC, Coutinho P. The global epidemiology of hereditary ataxia cerebellum in aging and neurodegenerative diseases. Neurobiol Learn Mem 2019. doi: and spastic paraplegia: a systematic review of prevalence studies. Neuroepidemiology 10.1016/j.nlm.2019.01.004. Epub 2019 Jan 7. 2014;42:174–183. 39. Nikolay R, Wiederkehr T, Rist W, Kramer G, Mayer MP, Bukau B. Dimerization of the 17. Pakdaman Y, Sanchez-Guixe M, Kleppe R, et al. In vitro characterization of six STUB1 human E3 ligase CHIP via a coiled-coil domain is essential for its activity. J Biol Chem variants in spinocerebellar ataxia 16 reveals altered structural properties for the 2004;279:2673–2678. encoded CHIP proteins. Biosci Rep 2017;37:BSR20170251. 40. Al-Ramahi I, Lam YC, Chen HK, et al. CHIP protects from the neurotoxicity of 18. Wang J, Al-Ouran R, Hu Y, et al. MARRVEL: integration of human and model expanded and wild-type ataxin-1 and promotes their ubiquitination and degradation. organism genetic resources to facilitate functional annotation of the human genome. J Biol Chem 2006;281:26714–26724. Am J Hum Genet 2017;100:843–853. 41. Scaglione KM, Zavodszky E, Todi SV, et al. Ube2w and ataxin-3 coordinately regulate 19. Kopanos C, Tsiolkas V, Kouris A, et al. VarSome: the human genomic variant search the ubiquitin ligase CHIP. Mol Cell 2011;43:599–612. engine. Bioinformatics 2018;35:1978–1980. 42. Dickey CA, Yue M, Lin WL, et al. Deletion of the ubiquitin ligase CHIP leads to the 20. Li J, Shi L, Zhang K, et al. VarCards: an integrated genetic and clinical database accumulation, but not the aggregation, of both endogenous phospho- and caspase-3- for coding variants in the human genome. Nucleic Acids Res 2018;46: cleaved tau species. J Neurosci 2006;26:6985–6996. D1039–D1048. 43. Wechsler D. WMS-R. Wechsler Memory Scale—Revised. San Antonio: Psychological 21. Turkgenc B, Sanlidag B, Eker A, et al. STUB1 polyadenylation signal variant AACAAA Corp: Harcourt Brace Jovanovich; 1987. does not affect polyadenylation but decreases STUB1 translation causing SCAR16. 44. King FW, Bird TD. Trail making test: norms for college males and relationship to Hum Mutat 2018;39:1344–1348. scholastic aptitude. Percept Mot Skills 1965;21:199–206. 22. Bar-Peled O, Ben-Hur H, Biegon A, et al. Distribution of glutamate transporter 45. Woodcock RW, Johnson B. Woodcock-Johnson Psycho-Educational Battery– subtypes during human brain development. J Neurochem 1997;69:2571–2580. Revised. Rolling Meadows, IL: Riverside; 1989. 23. Dehnes Y, Chaudhry FA, Ullensvang K, Lehre KP, Storm-Mathisen J, Danbolt NC. 46. Woodcock R, Johnson B. Woodcock-Johnson Psychoeducational Battery-Revised The glutamate transporter EAAT4 in rat cerebellar Purkinje cells: a glutamate-gated Tests of Achievement. Chicago: Riverside Publishing; 1990. chloride channel concentrated near the synapse in parts of the dendritic membrane 47. Sheslow D, Adams W. Wide Range Assessment of Memory and Learning (WRAML). facing astroglia. J Neurosci 1998;18:3606–3619. Wilmington, DE: Jastak; 1990. 24. Chen DH, Meneret A, Friedman JR, et al. ADCY5-related dyskinesia: broader 48. Grant DA, Berg E. A behavioral analysis of degree of reinforcement and ease of shifting to spectrum and genotype-phenotype correlations. Neurology 2015;85:2026–2035. new responses in Weigl-type card-sorting problem. J Exp Psychol 1948;38:404–411. 25. Barrett MJ, Williams ES, Chambers C, Dhamija R. Autosomal recessive inheritance of 49. Hooper H. Hooper Visual Organization Test (VOT). Los Angeles: Western Psy- ADCY5-related generalized dystonia and myoclonus. Neurol Genet 2017;3:193. chological Services; 1983. 26. Lise S, Clarkson Y, Perkins E, et al. Recessive mutations in SPTBN2 implicate beta-III 50. Wiens AN, Tindall AG, Crossen JR. California verbal learning test: a normative data spectrin in both cognitive and motor development. PLoS Genet 2012;8:e1003074. study. J Clin Neuropsychologist 1994;8:75–90.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 13 ARTICLE OPEN ACCESS ALS in Danish Registries Heritability and links to psychiatric and cardiovascular disorders

Betina B. Trabjerg, MSc, Fleur C. Garton, PhD, Wouter van Rheenen, MD, PhD, Fang Fang, MD, PhD, Correspondence Robert D. Henderson, MBBS, PhD, Preben Bo Mortensen, DrMedSc, Esben Agerbo, DrMedSc, and Dr. Wray [email protected]. Naomi R. Wray, PhD

Neurol Genet 2020;6:e398. doi:10.1212/NXG.0000000000000398 Abstract Objective To investigate the genetic contribution to amyotrophic lateral sclerosis (ALS) and the phenotypic and genetic associations between ALS and psychiatric and cardiovascular disorders (CVD) we used the national registry data from Denmark linked to first-degree relatives to estimate herita- bility and cross-trait parameters.

Methods ALS cases and 100 sex and birth-matched controls per case from the Danish Civil Registration System were linked to their records in the Danish National Patient Registry. Cases and controls were compared for (1) risk of ALS in first-degree relatives, used to estimate heritability, (2) comorbidity with psychiatric disorders and CVD, and (3) risk of psychiatric disorders and CVD in first-degree relatives.

Results 5,808 ALS cases and 580,800 controls were identified. Fifteen percent of cases and controls could be linked to both parents and full siblings, whereas 70% could be linked to children. (1) We estimated the heritability of ALS to be 0.43 (95% CI, 0.34–0.53). (2) We found increased rates of diagnosis of mental disorders (risk ratio = 1.18; 95% CI, 1.09–1.29) and CVD in those later diagnosed with ALS. (3) In first-degree relatives of those with ALS, we found increased rate of schizophrenia (1.17; 95% CI, 0.96–1.42), but no evidence for increased risk CVD.

Conclusions Heritability of ALS is lower than commonly reported. There is likely a genetic relationship between ALS and schizophrenia, and a nongenetic relationship between ALS and CVD.

From the National Centre for Register-Based Research NCRR (B.B.T., P.B.M., E.A.), Aarhus University; Centre for Integrated Register-Based Research CIRRAU (B.B.T., P.B.M., E.A.), Aarhus University; The Lundbeck Foundation Initiative for Integrative Psychiatric Research (B.B.T., P.B.M., E.A.), iPSYCH, Denmark; Institute for Molecular Bioscience (F.C.G., N.R.W.), University of Queensland, Brisbane, Australia; Department of Neurology, Brain Center Rudolf Magnus, University Medical Center Utrecht, The Netherlands; Department of Medical Epidemiology and Biostatistics (F.F.), Karolinska Institutet, Stockholm, Sweden; Centre for Clinical Research (R.D.H.), The University of Queensland, Brisbane; Queensland Brain Institute (R.D.H., N.R.W.), University of Queensland, Brisbane; Department of Neurology (R.D.H.), Royal Brisbane and Women’s Hospital, Australia.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing charge was funded by the authors.

The analyses based on the summary statistics derived from the Danish Registry data and quantitative genetic modeling was conducted by Fleur Garton and Naomi Wray. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary ALS = amyotrophic lateral sclerosis; CRS = Civil Registration System; CVD = cardiovascular disease/cardiovascular disorder; GWAS = genome-wide association study; ICD = International Classifications of Disease; MI = myocardial infarction; RR = risk ratio.

Multiple genes1 (e.g., SOD1 and C9orf72) can harbor causal January 1, 2017). The patient registry records any individual mutations of amyotrophic lateral sclerosis (ALS) found in with a hospital admission and covers information on all Danish 5%–10% of those diagnosed,2 mostly those with many affected somatic inpatient hospital contacts; outpatient contacts were family members. Genetic factors likely contribute to ALS when included from January 1, 1994. We considered both main and these mutations are absent. Heritability of liability quantifies auxiliary diagnoses for inpatient and outpatient contacts. In- the relative genetic contribution to the disease, assuming clusion was based on persons who were living in Denmark at – a polygenic architecture. Estimates of heritability3 7 (table 1) the time, were older than 16 years at the time of their first are difficult to obtain for late-onset disorders because the es- diagnosis, and had International Classifications of Disease sential information required is the increased risk of disease in (ICD) 8:348.0 or icd-10:G12.2 codes included in their records. relatives of those affected. This is particularly difficult for These definitions have been used in previous studies of ALS – a disorder such as ALS where lifetime risk is ;0.25%.8,9 using the Danish registry,19 22 and a very good agreement in diagnosis has been reported with death certificate data.19 A More recently, genome-wide association study (GWAS) data published validation study used these ICD codes for cases have enabled estimation of the proportion of variation in liability identified from the registry data and confirmed the ALS di- associated with genome-wide single nucleotide polymorphisms agnosis in 160 of 173 ALS cases (92.5%) and found no dif- (SNP-based heritability10). For ALS, SNP-based heritability has ference in the predictive validity between ICD-8 and ICD-10 been estimated to be 8.2% (95% CI, 7.2–9.2),11 which provides codes.22 a lower bound for heritability (because the estimate only captures variation associated with the GWAS common SNPs) and dem- Controls onstrates evidence in support of a polygenic architecture. SNP- For each case that was identified, 100 controls were extracted based heritability methods have bivariate extensions12,13 that allow and matched for sex and date of birth (alive and living in estimation of genetic correlations using GWAS summary statistics Denmark at the time case diagnosis was first recorded). Par- from independently collected disease cohorts. Application of these ticularly, for older cases, it was not possible to find 100 con- methods to ALS gave a significant estimate for a genetic corre- trols matched for an exact date of birth. Where necessary, the lation of ALS with schizophrenia (0.14, 95% CI, 0.07–0.21).14 By window was increased by ±1 day until the quota of controls contrast, the estimated genetic correlation with cardiovascular was reached. The controls were sampled without re- disease (CVD) was not different from zero,15 despite increased placement, and the cases could not be selected as controls. rates of CVD in those with ALS.16 Here, we apply genetic epi- demiology approaches to population records of ALS from Identification of family members Denmark with the objective to provide independent evidence for After the selection of ALS cases and controls, parents, siblings, genetic associations between ALS and other disorders. or children were identified when possible. For people regis- tered as children, a link to the CRS number of their legal parents is created. Since the registry was established in 1968, Methods only those born after ;1950 (aged 18 years or younger at the time when the registry was established) are expected to have Study population parental CRS records. Full siblings could only be identified for The Danish Civil Registration System (CRS) was established in those who had both parents linked to their records. Similarly, in 1968, registering all people alive and living in Denmark17 with identifying children of cases, only children born after 1950 and a unique personal identifier (CRS number). All other national before 2017 would be identified. We note that the CRS system registries can be linked using the CRS number. Our study identifies legal parents; although these are expected to mostly population included all persons alive in Denmark from January reflect biological relationships, this will not always be true. We 1, 1968, or born between 1969 and January 1, 2017. The study count parents, siblings, and children as first-degree relatives. was approved by the Danish Health Data Authority, the Danish Data Protection Agency, and Statistics Denmark and did not Psychiatric and CVD diagnoses require ethical approval because its use serves a scientificcause Information on individuals diagnosed with schizophrenia and and no reference is made to a single person. other mental disorders was obtained from the Danish Psychi- atric Central Research Register.23 From 1969, this Registry ALS cases contained data on all admissions to Danish psychiatric inpatient ALS cases were identified using the Danish National Patient facilities (secondary care treatment). From 1995, information Registry18 (established January 1, 1977, accessing records until on all contacts to outpatient psychiatric departments and visits

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 1 A summary of the studies estimating risk in relatives and/or estimating heritability

RR in first-degree Reported estimate of Reference Description Excludea Lifetime risk (95% CI) heritability (95% CI)

Graham et al.35 77 (26 MZ, 51DZ) twin pairs with at least No 0.38–0.85 one affected twin. 4 MZ concordant (2 from families with many affected individuals)

Fang et al.4 6,671 ALS cases identified in Swedish No 9.7 (7.2–12.8) national records. Prospective study following 1,906 siblings, 13,947 children, and 5,405 spouses

Al-Chalabi et al.3 171 (49 MZ, 122 DZ) twin pairs with at least No 0.25% 0.61 (0.38–0.78) one affected twin. 5 concordant MZ twin pairs. 0.19% to age 70 Includes Graham et al.35 sample.

Wingo et al.5 1,088 clinic probands, 8 had affected fathers No M: 0.20% Range: 2.2–6.9 0.36–0.48b and 16 had affected mothers. F: 0.15% Mid: 4.6

Hanby et al.36 1,502 clinic probands, 8/1,622 full siblings, Yes 0.3%c 8 18/1,545 children from the same clinic

Ryan et al.7 1,117 Irish registry probands. 18 parent/child Yes M: 0.29% 5.5 0.37 (0.20–0.54) affected pairs (after excluding C9orf72). F: 0.23% All parents recorded

This paper 5,808 Danish registry probands, 580,800 No 0.25%c 5.8 (4.17–8.00) 0.43 (0.34–0.53) sex-age matched controls

Abbreviations: MZ = monozygotic twins; DZ = dizygotic twins; RR = relative risk. a Here, we note if cases carrying known mutations or from families with many affected individuals were excluded, recognizing that any exclusion is likely to be incomplete. b The range reflects estimates based on alternative scenarios, particularly changing lifetime risk. c Assumed, not estimated in data. to psychiatric emergency care units was included. For all attributed to a polygenic contribution(thesameassumptionsas cases and controls and their identified first-degree relatives, made in the studies listed in table 1) and estimate the heritability – endorsements of records for psychiatric diagnoses were of ALS under the liability threshold model,26 28 assuming obtained24 (table e-1, links.lww.com/NXG/A224). The same a lifetime risk of 0.25%. Recognizing that families harboring the hospital registry that was used to identify ALS cases, the Danish known causal mutations cannot be identified from the registry National Patient Registry, was also used to identify all persons data, we conducted sensitivity analyses and re-estimated herita- with any diagnosis of heart disease. We defined heart disease bility removing 20% of counts of ALS concordant relative pairs. under both a broad definition of CVD and a narrower defini- Last, we estimated the risk of psychiatric disorders or CVD in tion of myocardial infarction (MI)20 (table e-1). The psychi- relatives of ALS cases and controls. We used bivariate liability atric or CVD diagnoses in both cases and their matched threshold methods to link risk in relatives to genetic controls were restricted to those recorded before the first re- correlation,27,28 assuming estimates of heritability for these traits cord of ALS for the case. We estimated risk of psychiatric made from Danish national registry data.28 disorders and CVD in ALS cases and controls and calculated risk ratios (RRs). Data availability Access to individual-level Denmark data is governed by Danish Quantitative genetic modeling authorities. These include the Danish Data Protection Agency, We estimated the increased risk of ALS in relatives of the cases the Danish Health Data Authority, and Statistics Denmark. ff compared with the controls for di erent types of relatives. Each scientific project must be approved before initiation, and approval is granted to a specific Danish research institution. ff fi When families have 3 or more a ected rst-degree relatives, it is International researchers may gain data access through col- very likely that they have a mendelian type of ALS, given a life- laboration with a Danish research institution. time risk for ALS of ;0.25%. However, co-occurrence of ALS in a pair of relatives could be explained by shared genetic factors under a more polygenic model of disease (or even by shared Results environment). Many other diseases have mendelian and poly- genic types (such as breast cancer, Parkinson disease, and Alz- Cases and controls heimer disease). Under a polygenic genetic architecture, even We identified 5,808 ALS cases (3,227 men and 2,581 women), when the total genetic contribution to disease etiology is high, of which 62% had an ICD-10:G122 diagnosis, 27% ICD-8: most people affected present without a family history.25 Here, we 34809, and 10% ICD-10:G122G. These statistics are in good assume that the observed increased risk in first-degree relatives is agreement with the studies which previously accessed the ALS

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 – records from Danish registers.19 22 For more than 94% of cases, be linked to their mother, 15% could be linked to their all their matched controls were born within 2 days (table e-2, father, and 15% could be linked to both parents, with no links.lww.com/NXG/A224). The birth years of the cases ranged difference between the cases and controls (χ2 test, p = 0.12) from 1886 to 2000, but given that ALS is typically a disease of (table e-4, links.lww.com/NXG/A224). Only those linked late-onset, 94% of cases were born before 1960 (table e-3). We to their parents could have their siblings identified (table e-5). found that annual incidence was higher in later years (figure e-1), By contrast, children could be identified for 70% of people consistent with the previously reported increase in age-adjusted (table e-6). Overall, 74% of people had at least one first-degree incidence of 1.6% annually after 1982.21 Because some people relative identified in the registry data (table e-7). In these with ALS have no need for an inpatient hospitalization, the data, no ALS case or control had more than one affected sibling addition of outpatient records in 1994 may partially account for and the number of individuals with more than one affected incidence increase. The median age at diagnosis for the ALS child was less than 4 (concern about potential reidentification cohort (minimum age of diagnosis of 16 years) was 67.1 years means that the exact number is nonreportable). The rate of (interquartile range 58.4–74.2 years) (figure e-2). ALS was 0.15% in mothers of controls and 0.20% in fathers of controls (among only the 15% controls with their parents Relatives identified) (table 2). Because ALS is a late-onset disorder, As a reflection of the timing of the registry establishment, the the parents of controls matched to the cases must be close to percentage of those with a link to a mother increased rapidly a full lifetime in age; hence, these rates provide a lower bound from 9.7% for those born in 1950 to 98.6% and 99.8% for on lifetime risk. those born in 1960 and 1970, respectively, retaining that level for all subsequent birth years. Similar findings were Estimate of heritability obtained regarding a link to a father. Given the age distri- The RR of ALS in first-degree relatives of ALS cases compared bution for ALS diagnosis, only 16% of cases/controls could with those of the controls ranged from 4.2 in fathers to 12.8 in

Table 2 Counts of ALS in each relative type of ALS cases and controls and RRs for ALS cases compared with ALS controls

No record of ALS in Relative Relative is recorded as having ALS

Control Case Total Control Case Total RRa

Mother

Count 95,224 895 96,119 144 8 152 5.87

%b 99.8 99.1 0.2 0.9 2.89–11.9

Father

Count 90,216 847 91,063 177 7 184 4.19

%b 99.8 99.2 0.2 0.8 1.97–8.88

Full sibling

Count 61,962 587 62,549 65 8 73 12.8

%b 99.9 98.7 0.1 1.3 6.2–26.6

Child

Count 414,290 4,050 418,340 277 15 292 5.52

%b 99.93 99.63 0.07 0.37 3.29–9.28

First-degree

Count 433,005 4,270 437,275 662 38 700 5.78

%b 99.8 99.1 0.2 0.9 4.17–8.00

Sensitivity first-degreec

Count 433,005 4,270 437,275 662 30 700 4.57

%b 99.8 99.1 0.2 0.7 3.17–6.58

Abbreviations: ALS = amyotrophic lateral sclerosis; RR = risk ratio. Bold type face is used for RR that are significantly different from 1 a Calculated function RR from R library fmsb. b Percent of either case or control count. c Sensitivity analyses exclude 20% of case-concordant pairs as possible monogenic ALS hence reducing the count of 38–30.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 3 Estimates of heritability of ALS based on relative risk of ALS for different type of relatives

Mothers Fathers Siblings Children First-degree Sensitivitya

Risk for relatives of controls (%) 0.15 0.20 0.10 0.07 0.15 0.15

Risk for relatives of cases (%) 0.89 0.82 1.34 0.37 0.88 0.70

Increased risk of ALS in relatives of those with ALS 5.87 4.19 12.83 5.52 5.78 4.57 s.e. of estimate 0.36 0.39 0.38 0.27 0.17 0.19

95% CI 2.89–11.9 1.97–8.88 6.18–26.6 3.29–9.28 4.17–8.00 3.17–6.58

Estimate of heritability, assuming lifetime risk 0.25% 0.44 0.34 0.68 0.42 0.43 0.36 s.e. of estimate 0.11 0.10 0.13 0.08 0.05 0.05

95% CI 0.24–0.66 0.15–0.56 0.45–0.95 0.28–0.58 0.34–0.53 0.27–0.47

Abbreviations: ALS = amyotrophic lateral sclerosis; RR = risk ratio; s.e. = standard error. – Heritability following liability threshold methodology,26 28 assuming lifetime risk of K = 0.0025. 95% CI are calculated using the 95% CI of the RR, as the only estimated value in the equation. a Sensitivity analysis using the risk to relatives from the sensitivity row in table 2. siblings (table 3). Despite this range, the 95% CI was over- 0.19). We also found an increased risk of all psychiatric disorders − lapping. Across all first-degree relatives, there is a 5.78-fold (RR = 1.18; 95% CI, 1.09–1.29, p = 1.3 × 10 4) likely reflecting (95% CI, 4.17–8.00) increase in the risk of ALS in relatives of an increased risk in affective disorders excluding bipolar disorder the cases compared with in relatives of the controls. This (RR = 1.20; 95% CI, 1.03–1.40, p = 0.019). These affective generates an estimate of heritability of 0.43 (95% CI, disorders affect 2.4% of the controls (i.e., reflecting mostly hos- 0.34–0.53), given a lifetime risk8,9 of 0.25% (table 3). Sensi- pitalized unipolar disorder), noting that the Danish Psychiatric tivity analyses assumed lifetime risks of ALS in Denmark of Central Research Register does not include depression treated in 0.2% or 0.3%, which gave heritability estimates of 0.41 (95% CI, primary care. There is no increased risk of severe psychiatric 0.32–0.50) and 0.45 (95% CI, 0.35–0.55), respectively, (tables disorders such as schizophrenia, bipolar disorder, and anorexia. e-8 and e-9, links.lww.com/NXG/A224). The methodology to estimate heritability assumes that the disease is polygenic, and We estimated the risk of psychiatric and CVD disorders in first- hence, the estimate provided here is an upper limit of herita- degree relatives of those with and without ALS. When con- bility on the liability scale because some of the first-degree sidering all first-degree relatives together, none of the risks to relatives concordant for ALS may reflect families harboring relatives were significant (table 5). The highest increased risk mendelian mutations; as such, it is not possible to identify these was 1.17 (95% CI, 0.97–1.42; p = 0.10) for schizophrenia (table families from the registry data. In a sensitivity analysis, with an 5). Despite being nonsignificant, the increased risk was in the exclusion of 20% of concordant pairs, the estimate of the rel- direction expected from the genetic correlation estimates from ative risk reduced to 4.57 (95% CI, 3.17–6.58) and that of GWAS summary data.14 Using lifetime risk and heritability heritability reduced to 0.36 (95% CI, 0.27–0.47) (table 3). We estimates28 0.0112 and 0.67, respectively for schizophrenia, and note that no family included more than 2 affected individuals, of 0.0025 and 0.43, respectively for ALS, and assuming a ge- and we were unable to find reports of rates of ALS mendelian netic correlation of 0.14,14 we estimate the expected RR to be mutations in Denmark. Given the intercountry variability, it is 1.32, which is within the 95% CI of our estimate. When con- possible that the rates of ALS dominant mutations are not high. sidering different first-degree relatives, we found an increased For example, in an international study,29 the most common risk of schizophrenia in mothers of ALS cases (RR = 2.21; 95% form of mendelian ALS, associated with the hexanucleotide CI, 1.05–4.65, p = 0.033) and increased risk of all mental repeat in the gene C9orf72, was less frequent in the Netherlands disorders in fathers of ALS case (RR = 1.29; 95% CI, 1.08–1.53, and Sweden (albeit the latter from a small sample) than in p = 0.0053) (table e-8, links.lww.com/NXG/A224). other countries; Denmark did not contribute to the study. Discussion Cross-trait analyses within individual and for relatives Recorded in over 48 years of Danish national patient and We estimated the risk of psychiatric disorders and CVD before outpatient registries were 5,808 ALS cases, who were matched ALS diagnosis date in those with ALS and their matched controls with 580,800 controls. The increased risk of ALS in first- (table 4). We found an increased risk of CVD (RR) of (1.21; 95% degree relatives of those with ALS compared with first-degree − CI, 1.15–1.27, p =6×10 15) in those with ALS compared with relatives of controls was 5.78 (95% CI, 4.17–8.00). The those without (as previously reported16), but the increased risk resulting estimate of heritability was 0.43 (95% CI, 0.34–0.53). was not significant for MI (RR = 1.08; 95% CI, 0.96–1.22, p = This has a lower mean and smaller bounds than the published

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Table 4 Disease/disorder counts in 5,808 ALS cases and 580,800 matched controls before ALS diagnosis in the case and RRs for ALS cases compared with ALS controls

Given disorder not recorded for individual Given disorder is recorded for individual

Control Case Total Control Case Total RR 95% CI

All mental disorders

Count 539,719 5,322 545,041 41,081 486 41,567 1.18

%a 92.9 91.6 7.1 8.4 1.09–1.29

Schizophrenia

Count 578,125 5,785 583,910 2,675 23 2,698 0.86

%a 99.5 99.6 0.5 0.4 0.57–1.30

Other psychotic disorders

Count 576,735 5,766 582,501 4,065 42 4,107 1.03

%a 99.3 99.3 0.7 0.7 0.76–1.40

Bipolar disorder

Count 578,159 5,772 583,931 2,641 36 2,677 1.36

%a 99.6 99.4 0.5 0.6 0.98–1.89

Other affective disorders

Count 567,121 5,644 572,765 13,679 164 13,843 1.20

%a 97.6 97.2 2.4 2.8 1.03–1.40

Anorexia

Count 580,719 5,808 586,527 81 0 81 0.62

%a 99.99 100.00 0.01 0.00 0.04–10.0

Heart disease

Count 467,227 4,435 471662 113,573 1,373 114,946 1.21

%a 80.5 76.4 19.6 23.6 1.15–1.27

MI

Count 555,650 5,536 561,186 25,150 272 25,422 1.08

%a 95.7 95.3 4.3 4.7 0.96–1.22

Abbreviations: ALS = amyotrophic lateral sclerosis; MI = myocardial infarction; RR = risk ratio. Note 0.5 was added to the case count for anorexia to enable estimation of the RR. Bold type face is used for RR significantly different from 1. a Percent of either case or control count.

twin-based estimate3 (0.61%, 95% CI, 0.38–0.78%) but is in with schizophrenia: reported from survival analysis using the good agreement with later estimates (table 1). Danish registry data to be 18.7 years and 16.3 years shorter for men and women, respectively.30 The observed increased rate of The published estimate of the genetic correlation derived from schizophrenia in first-degree relatives of those with ALS was ALS and schizophrenia GWAS data is 0.14 (95% CI, also not significant (1.17; 95% CI, 0.96–1.42; p = 0.10), but the 0.07–0.21).14 Although psychosis has been reported in relatives RR expected from a genetic correlation of 0.14 is 1.32, which is of those with ALS in kindred segregating the C9orf72 hex- within the confidence limits of our estimate. Hence, our results anucleotide repeat expansion,2 increased rates of ALS in those provide some support that the genetic relationship between with schizophrenia have not been reported, which would be ALS and schizophrenia is genetic. A Swedish national registry expected if there was a shared genetic architecture between the study, which identified 3,648 ALS cases and 364,800 matched disorders. Here, we found no evidence for increased rates of controls found higher rates of psychiatric disorders in chil- schizophrenia in those with ALS (RR 0.86, 95% CI, 0.57–1.30). dren of those with ALS.31 The overlap between ALS and However, the life expectancy is significantly reduced for those schizophrenia is unlikely to be attributed to C9orf72 mutations

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 5 Disease/disorder counts in first-degree relatives of ALS cases and controls, and RRs for ALS cases compared to ALS controls

No first-degree relatives with One or more first-degree relatives with the given disorder the given disorder

Control Case Total Control Case Total RR 95% CI

ALS

Count 433,005 4,270 437,275 662 38 700 5.78

%a 99.8 99.1 0.2 0.9 4.17–8.00

All mental disorders

Count 321,560 3,169 324,729 110,622 1,121 111,743 1.02

%a 74.4 73.9 25.6 26.1 0.97–1.07

Schizophrenia

Count 423,146 4,185 427,331 9,036 105 9,141 1.17

%a 97.9 97.6 2.1 2.4 0.97–1.42

Other psychotic disorders

Count 422,172 4,188 426,360 10,010 102 10,112 1.03

%a 97.7 97.6 2.3 2.4 0.85–1.24

Bipolar disorder

Count 424,690 4,215 428,905 7,492 75 7,567 1.01

%a 98.3 98.3 1.7 1.7 0.80–1.26

Other affective disorders

Count 393,344 3,915 397,259 38,838 375 39,213 0.97

%a 91.0 91.3 9.0 8.7 0.88–1.07

Anorexia

Count 430,662 4,277 434,939 1,520 13 1,533 0.86

%a 99.6 99.7 0.4 0.3 0.50–1.49

Heart disease

Count 278,665 2,832 281,497 153,517 1,458 154,975 0.96

%a 64.5 66.0 35.5 34.0 0.92–1.00

MI

Count 398,155 3,984 402,139 34,027 306 34,333 0.91

%a 92.1 92.9 7.9 7.1 0.81–1.01

Abbreviations: ALS = amyotrophic lateral sclerosis; MI = myocardial infarction; RR = risk ratio. Bold type face is used for RR significantly different from 1. Note: Only persons with recorded first-degree relatives are included. First-degree relatives accounting parents, full siblings, and children. a Percent of either case or control count. because the hexanucleotide repeat is reported to have a fre- among relatives of those with ALS, despite high rates in first- quency less than 0.1% in those with schizophrenia.32 degree relatives of both the cases and controls (;35%). This may suggest broad lifestyle factors, rather than a genetic re- We found increased rates of recorded cardiovascular disease lationship, and explain the observed increased rates of CVD in (CVD) before diagnosis of ALS (RR = 1.21; 95% CI, those with ALS,16 which would be consistent with the genetic 1.15–1.27), as previously reported.16 This association, how- correlation estimates from GWAS data between ALS and CVD ever, was not detected in the narrow definition of heart disease; that were not different from zero.15 Alternatively, the high rate MI. By contrast, we found no increase in the rate of CVD of CVD recorded before ALS diagnosis may reflect the

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 increased medical surveillance in those in the early stages of financial support from The Stanley Medical Research Institute disease. and The Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH (grant no. R102-A9118 and The key strength of the Danish registry data is its completeness R155-2014-1724). so that ascertainment biases are minimized compared with other designs. Nonetheless, there are limitations. First, because Study funding ALS is a late-onset disorder and given the timing of the es- The Australian National Health and Medical Research tablishment of the registry system, we could only trace parents Council (1078901, 1113400, 1087889) to N.R.W. and F.G. for those born after ;1950. From the parents, we were able to The Stanley Medical Research Institute and The Lundbeck trace the siblings, but only siblings born after 1950. To counter Foundation Initiative for Integrative Psychiatric Research, this, our design of 100 controls per case provides robustness to iPSYCH (grant no. R102-A9118 and R155-2014-1724) to our estimates and any potential biases apply to both cases and B.B.T., P.B.M., and E.A. controls. Second, although ;70% of ALS cases had records of first-degree relatives in the registry system, the majority of such Disclosure relatives were children, who were mostly below the expected B.B. Trabjerg, F.C. Garton, W. van Rheenen, F. Fang, R.D. age of ALS onset (median age diagnosis in these data of 67.1 Henderson, P.B. Mortensen, E. Agerbo, and N.R. Wray report years). The information for estimation of heritability was no disclosures. Go to Neurology.org/NG for full disclosures. therefore mostly generated from ;15% of the ALS cases that could be linked to both parents and hence also to siblings, Publication history which is still a sizeable data set (table 1). Third, in these registry Received by Neurology: Genetics September 10, 2019. Accepted in final data, we were unable to identify individuals and their family form January 13, 2020. members carrying ALS causal mutations and inclusion of these individuals could inflate the estimate of heritability, which methodologically assumes a polygenic architecture. We did not Appendix Authors

attempt to access frontotemporal dementia diagnoses (relevant Name Location Role Contribution because it co-occurs in families carrying the C9orf72 hex- 33 Betina B. Aarhus Author Analysis of the Danish anucleotide repeat ) because these diagnoses have not been Trabjerg, University, registry data 34 validated in registry data. To address these limitations, we MSc Denmark conducted sensitivity analyses in which we excluded from the Fleur C. The University of Author Analysis applied to the analyses a proportion of concordant relative pairs, the pro- Garton, PhD Queensland, summary statistics portion chosen as a likely upper limit that could be attributed to Brisbane, derived from the Danish Australia registry data. Manuscript first-degree relative pairs with causal mutations. The resulting first draft. relative risk (RR = 4.57; 95% CI, 3.17–6.58) and heritability – Wouter van University Author Study design and ALS (RR = 0.36; 95% CI, 0.27 0.47) provide lower bounds of these Rheenen, medical Center interpretation estimates. Last, higher relative risks associated with maternal MD, PhD Utrecht, The Netherlands transmission have recently been reported in a prospective ALS study in Ireland,7 although results were not significant when Fang Fang, Karolinska Author Study design and MD, PhD Institutet, interpretation families carrying the C9orf72 hexanucleotide repeat expansion Stockholm, were excluded. Here, we have not considered sex-specificrisks Sweden ff because the number of concordant a ected pairs is too few to Robert D. The University of Author Study design and ALS generate meaningful results. Henderson, Queensland, interpretation MBBS, PhD Brisbane, Australia Our heritability estimate of 0.43 withstands sensitivity analyses and supports a genetic contribution to the disease. If anything, Preben Bo Aarhus Author Supervision of the Mortensen, University, analysis of Danish registry our estimate might be upwardly biased, yet the estimate is lower DrMedSc Denmark data; study design than the twin estimate of heritability of ALS that has been most Esben Aarhus Author Supervision of the often cited. The weight of evidence is now such that a lower Agerbo, University, analysis of Danish registry estimate must be seen as more plausible (table 1), we suggest DrMedSc Denmark data; study design ; fi 0.40. We nd evidence for previous risk of psychiatric dis- Naomi R. The University of Author Study design. Analysis orders and CVD in individuals with ALS, plus tentative evi- Wray, PhD Queensland, applied to the summary Brisbane, statistics derived from the dence that this relationship has a genetic contribution for Australia Danish registry data. schizophrenia and other psychiatric disorders but not for CVD. Manuscript first draft.

Acknowledgment N.R.W. and F.C.G. acknowledge funding from the Australian References National Health and Medical Research Council (1078901, 1. Taylor JP, Brown RH Jr, Cleveland DW. Decoding ALS: from genes to mechanism. 1113400, 1087889). B.B.T., P.B.M., and E.A. acknowledge Nature 2016;539:197–206.

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 2. Byrne S, Walsh C, Lynch C, et al. Rate of familial amyotrophic lateral sclerosis: 20. Laursen TM, Munk-Olsen T, Agerbo E, Gasse C, Mortensen PB. Somatic hospital a systematic review and meta-analysis. J Neurol Neurosurg Psychiatry 2011;82: contacts, invasive cardiac procedures, and mortality from heart disease in patients with 623–627. severe mental disorder. Arch Gen Psychiatry 2009;66:713–720. 3. Al-Chalabi A, Fang F, Hanby MF, et al. An estimate of amyotrophic lateral sclerosis 21. Seals RM, Hansen J, Gredal O, Weisskopf MG. Age-period-cohort analysis of trends heritability using twin data. J Neurol Neurosurg Psychiatry 2010;81:1324–1326. in amyotrophic lateral sclerosis in Denmark, 1970–2009. Am J Epidemiol 2013;178: 4. Fang F, Kamel F, Lichtenstein P, et al. Familial aggregation of amyotrophic lateral 1265–1271. sclerosis. Ann Neurol 2009;66:94–99. 22. Seals RM, Kioumourtzoglou MA, Hansen J, Gredal O, Weisskopf MG. Amyotrophic 5. Wingo TS, Cutler DJ, Yarab N, Kelly CM, Glass JD. The heritability of amyotrophic lateral sclerosis and the Military: a population-based study in the Danish Registries. lateral sclerosis in a clinically ascertained United States research registry. PLoS One Epidemiology 2016;27:188–193. 2011;6:e27985. 23. Mors O, Perto GP, Mortensen PB. The Danish psychiatric Central research register. 6. Hanby MF, Scott KM, Scotton W, et al. The risk to relatives of patients with sporadic Scand J Public Health 2011;39:54–57. amyotrophic lateral sclerosis. Brain 2011;134:3454–3457. 24. Pedersen CB, Mors O, Bertelsen A, et al. A Comprehensive nationwide study of the 7. Ryan M, Heverin M, McLaughlin RL, Hardiman O. Lifetime risk and heritability of incidence rate and lifetime risk for treated mental disorders. JAMA Psychiatry 2014; amyotrophic lateral sclerosis. JAMA Neurol 2019;76:1367–1374. 71:573–581. 8. Alonso A, Logroscino G, Jick SS, Hernan MA. Incidence and lifetime risk of motor neuron 25. Yang J, Visscher PM, Wray NR. Sporadic cases are the norm for complex disease. Eur J disease in the United Kingdom: a population-based study. Eur J Neurol 2009;16:745–751. Hum Genet 2010;18:1039–1043. 9. Kiernan MC, Vucic S, Cheah BC, et al. Amyotrophic lateral sclerosis. Lancet 2011; 26. Reich T, Morris CA, James JW. Use of multiple thresholds in determining mode of 377:942–955. transmission of semi-continuous traits. Ann Hum Genet 1972;36:163–184. 10. Yang J, Zeng J, Goddard ME, Wray NR, Visscher PM. Concepts, estimation and 27. Falconer DS. Inheritance of liability to certain diseases estimated from incidence interpretation of SNP-based heritability. Nat Genet 2017;49:1304. among relatives. Ann Hum Genet 1965;29:51-76. 11. van Rheenen W, Shatunov A, Dekker AM, et al. Genome-wide association analyses 28. Wray NR, Gottesman II. Using summary data from the Danish national registers to identify new risk variants and the genetic architecture of amyotrophic lateral sclerosis. estimate heritabilities for schizophrenia, bipolar disorder, and major depressive dis- Nat Genet 2016;48:1043–1048. order. Front Genet 2012;3:118. 12. Lee SH, Yang J, Goddard ME, Visscher PM, Wray NR. Estimation of pleiotropy 29. Majounie E, Renton AE, Mok K, et al. Frequency of the C9orf72 hexanucleotide between complex diseases using single-nucleotide polymorphism-derived genomic repeat expansion in patients with amyotrophic lateral sclerosis and frontotemporal relationships and restricted maximum likelihood. Bioinformatics 2012;28:2540–2542. dementia: a cross-sectional study. Lancet Neurol 2012;11:323–330. 13. Bulik-Sullivan B, Finucane HK, Anttila V, et al. An atlas of genetic correlations across 30. Laursen TM. Life expectancy among persons with schizophrenia or bipolar affective human diseases and traits. Nat Genet 2015;47:1236–1241. disorder. Schizophr Res 2011;131:101–104. 14. McLaughlin RL, Schijven D, van Rheenen W, et al. Genetic correlation between 31. LonginettiE,MariosaD,LarssonH,etal. Neurodegenerative and psychiatric amyotrophic lateral sclerosis and schizophrenia. Nat Commun 2017;8:14774. diseases among families with amyotrophic lateral sclerosis. Neurology 2017;89: 15. Bandres-Ciga S, Noyce AJ, Hemani G, et al. Shared polygenic risk and causal infer- 578–585. ences in amyotrophic lateral sclerosis. Ann Neurol 2019;85:470–481. 32. Ducharme S, Bajestan S, Dickerson BC, Voon V. Psychiatric Presentations of C9orf72 16. Kioumourtzoglou MA, Seals RM, Gredal O, Mittleman MA, Hansen J, Weisskopf mutation: what are the diagnostic implications for Clinicians? J Neuropsychiatry Clin MG. Cardiovascular disease and diagnosis of amyotrophic lateral sclerosis: a pop- Neurosci 2017;29:195–205. ulation based study. Amyotroph Lateral Scler Frontotemporal Degeneration 2016;17: 33. Rohrer JD, Isaacs AM, Mizielinska S, et al. C9orf72 expansions in fronto- 548–554. temporal dementia and amyotrophic lateral sclerosis. Lancet Neurol 2015;14: 17. Pedersen CB. The Danish Civil Registration system. Scand J Public Health 2011;39: 291–301. 22–25. 34. Phung TK, Andersen BB, Hogh P, Kessing LV, Mortensen PB, Waldemar G. Validity 18. Lynge E, Sandegaard JL, Rebolj M. The Danish national patient register. Scand J of dementia diagnoses in the Danish hospital registers. Dement Geriatr Cogn Disord Public Health 2011;39:30–33. 2007;24:220–228. 19. Kioumourtzoglou MA, Seals RM, Himmerslev L, Gredal O, Hansen J, Weisskopf MG. 35. Graham AJ, Macdonald AM, Hawkes CH. British motor neuron disease twin study. Comparison of diagnoses of amyotrophic lateral sclerosis by use of death certificates J Neurol Neurosurg Psychiatry 1997;62:562–569. and hospital discharge data in the Danish population. Amyotroph Lateral Scler 36. Hanby MF, Scott KM, Scotton W, et al. The risk to relatives of patients with sporadic Frontotemporal Degener 2015;16:224–229. amyotrophic lateral sclerosis. Brain 2011;134:3454–3457.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 ARTICLE OPEN ACCESS MYORG-related disease is associated with central pontine calcifications and atypical parkinsonism

Viorica Chelban, MD, MSc, MRCP, Miryam Carecchio, MD, PhD, Gillian Rea, MD, Abdalla Bowirrat, MD, PhD, Correspondence Salman Kirmani, MD, Luca Magistrelli, MD, Stephanie Efthymiou, MSc, Lucia Schottlaender, MD, PhD, Dr. Chelban [email protected] Jana Vandrovcova, PhD, Vincenzo Salpietro, MD, Ettore Salsano, MD, Davide Pareyson, MD, or Dr. Garavaglia Luisa Chiapparini, MD, Farida Jan, MD, Shahnaz Ibrahim, MD Prof, Fatima Khan, MD, Zul Qarnain, MD, [email protected] Stanislav Groppa, MD, PhD, Nin Bajaj, MD, PhD, Bettina Balint, MD, Kailash P. Bhatia, MD Prof, Andrew Lees, FMedSci, FRCP, Patrick J. Morrison, CBE, MD, DSc, Nicholas W. Wood, MD, PhD, Barbara Garavaglia, MD, PhD, and Henry Houlden, MD, PhD

Neurol Genet 2020;6:e399. doi:10.1212/NXG.0000000000000399 Abstract Objective To identify the phenotypic, neuroimaging, and genotype-phenotype expression of MYORG mutations.

Methods Using next-generation sequencing, we screened 86 patients with primary familial brain calci- fication (PFBC) from 60 families with autosomal recessive or absent family history that were negative for mutations in SLC20A2, PDGFRB, PDGBB, and XPR1. In-depth phenotyping and neuroimaging investigations were performed in all cases reported here.

Results We identified 12 distinct deleterious MYORG variants in 7 of the 60 families with PFBC. Overall, biallelic MYORG mutations accounted for 11.6% of PFBC families in our cohort. A heteroge- neous phenotypic expression was identified within and between families with a median age at onset of 56.4 years, a variable combination of parkinsonism, cerebellar signs, and cognitive decline. Psychiatric disturbances were not a prominent feature. Cognitive assessment showed impaired cognitive function in 62.5% of cases. Parkinsonism associated with vertical nuclear gaze palsy was the initial clinical presentation in 1/3 of cases and was associated with central pontine calcifications. Cerebral cortical atrophy was present in 37% of cases.

Conclusions This large, multicentric study shows that biallelic MYORG mutations represent a significant proportion of autosomal recessive PFBC. We recommend screening MYORG mutations in all patients with primary brain calcifications and autosomal recessive or negative family history, especially when presenting clinically as atypical parkinsonism and with pontine calcification on brain CT.

From the Department of Neuromuscular Diseases (V.C., S.E., L.S., J.V., V.S., N.W.W., H.H.), UCL Queen Square Institute of Neurology; National Hospital for Neurology and Neurosurgery (V.C., S.E., L.S., J.V., V.S., N.W.W., H.H.), Queen Square, London, UK; Department of Neurology and Neurosurgery (V.C., S.G.), Institute of Emergency Medicine, Chisinau, Republic of Moldova; Department of Neuroscience (M.C.), University of Padua, Italy; Northern Ireland Regional Genetics Service (G.R., P.J.M.), Belfast City Hospital, UK; Department of Neuroscience (A.B.), Interdisciplinary Center (IDC) Herzliya, Israel; Department of Paediatrics & Child Health (S.K., F.J., S.I., F.K., Z.Q.), Aga Khan University, Karachi, Pakistan; Department of Neurology (L.M.), Eastern Piedmont University, Novara, Italy; Department of Neurology (E.S., D.P.) and Department of Neuroradiology (L.C.), Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy; Department of Clinical Neurology (N.B.), University of Nottingham, UK; Department of Clinical and Movement Neurosciences (B.B., K.P.B., N.W.W.), UCL Queen Square Institute of Neurology, London, UK; Department of Neurology (B.B.), Heidelberg University Hospital, Germany; Reta Lila Weston Institute (A.L.), UCL Queen Square Institute of Neurology, London, UK; and Medical Genetics and Neurogenetics Unit (B.G.), Fondazione IRCCS Istituto Neurologico Carlo Besta, Milan, Italy.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary DAT = dopamine active transporter; gnomAD = Genome Aggregation Database; MMSE = Mini-Mental State Examination; PFBC = primary familial brain calcification.

Primary familial brain calcification (PFBC) is a genetic neu- 100 bp paired-end reads. Alignment was performed using BWA rodegenerative condition characterized by calcium deposition (bio-bwa.sourceforge.net/)7 with GRCH38 as a reference. – in the basal ganglia and other brain regions usually presenting Variants were called using the GATK8 11 UnifiedGenotyper- – with a combination of movement disorders, migraine, psychi- based pipeline8 10 workflow. All variants were annotated using atric, and cognitive impairment. The exact prevalence of PFBC ANNOVAR12 and filtered using custom R scripts. Only novel is unknown, but population-based genomic analysis indicates or very rare variants with a minor allele frequency of <0.01 in that it is underestimated and underdiagnosed,1 with a molecu- the 1000 Genomes Project13 and Genome Aggregation Data- lar diagnosis achieved in only up to 50% of PFBC cases.2 The base (gnomAD)14 or known pathologic mutations were in- pathogenesis of PFBC involves calcium and phosphate ho- cluded. Variants were filtered for homozygous, compound meostasis via mutations in SLC20A2 (OMIM: 158378) and heterozygous, highly deleterious, rare mutations segregating XPR1 (OMIM: 605237) and endothelial integrity and function with the disease. Except for families 1 and 7, segregation was affecting the blood-brain barrier via mutations in PDGFB confirmed in all other families. (OMIM: 190040) and PDGFRB (OMIM: 173410). Among these, mutations in SLC20A2 account for approximately 45% of For every rare MYORG variant identified (ENST00000297625, all autosomal dominant and de novo reported familial cases GenBank transcript ID NM_020702), we determined patho- from diverse ethnicities.3 However, a large proportion of au- genicity and novelty. Pathogenicity was assessed using the tosomal recessive PFBC remain undiagnosed.4 Recently, bial- American College of Medical Genetics and Genomics and the lelic mutations in MYORG (OMIM: 618255) have been Association for Molecular Pathology recommendations for implicated in the pathogenesis of autosomal recessive PFBC in variant classification.15 Only pathogenic and likely pathogenic families of Chinese5 and French6 ethnicity. Here, we report variants were included here. All pathogenic and likely patho- a large multicentric cohort of ethnically diverse patients with genic variants were confirmed with bidirectional Sanger se- biallelic variants in MYORG and broaden the phenotypic quencing. Primers are available in table e-1 (links.lww.com/ spectrum related to MYORG mutations. NXG/A227).

Standard protocol approvals, registrations, Methods and patient consents The individuals included in this study were recruited along Patients with unaffected family members under ethics-approved re- Patients with an autosomal recessive or negative family history search protocols (UCLH: 04/N034) with informed consent. and confirmed clinical and radiologic diagnosis of PFBC were recruited from multiple centers. Genetic testing was performed Data availability on stored blood samples of patients with unidentified etiologies Anonymized data used for this study are available from the of PFBC. Ethnically, the families were of British, Italian, Irish, corresponding authors on reasonable request. A data access Pakistani, and Israeli origin. Secondary causes of brain calcifi- agreement needs to be signed. cation were excluded in all cases. All cases were negative for other PFBC-related genes (SLC20A2, PDGFRB, PDGBB, and XPR1) and had comprehensive phenotyping performed by Results neurogenetics specialists. Genetic spectrum In cases with biallelic MYORG variants, the results from addi- We screened 86 cases from 60 families with PFBC that were tional investigations were retrospectively analyzed based on negative for pathogenic variants in SLC20A2, PDGFB, chart review where available: neuroimaging with CT in all PDGFRB,andXPR1 and had a recessive or negative family reported cases (n = 8), brain MRI (n = 4), dopamine active history. We identified pathogenic and likely pathogenic ho- transporter (DAT) scan (n = 2), and fluorodeoxyglucose-PET mozygous and compound heterozygous variants in MYORG (n = 2). Cognitive impairment was assessed by formal (ENST00000297625, GenBank transcript ID NM_020702) in psychometry. 8 cases from 7 families (figure e-1, links.lww.com/NXG/ A227). Overall, biallelic MYORG mutations accounted for Genetic testing 11.6% (7/60) of PFBC families in our cohort. We identified 12 DNA was extracted from peripheral blood. Whole-exome se- distinct mutations, of which 4 were novel (figure 1A) and 8 quencing was performed in all families. An Illumina HiSeq4000 were present in gnomAD with very low allele frequency in the instrument (Illumina, San Diego, CA) was used to generate heterozygous state and absent in the homozygous state (table

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 1 Genetic spectrum of MYORG mutations

(A) Schematic representation of MYORG with all the mutations identified in our study and those reported to date. The MYORG functional domains and their cellular localization are indicated: green, N-terminal sequence (cytoplasmic); gray, helical (transmembrane); orange, glycosidase domain (extracellular). The mutations reported in this cohort are plotted on top of the gene; mutations previously reported are below the gene. (B) Conservation across species for novel missense MYORG variants. The variants are marked with red boxes for the corresponding amino acid. e-2). With the exception of 1 variant (p.Ile656Thr), none of the Insidious onset with brain calcifications found incidentally was variants presented here have been previously reported in in 12.5% of cases. Table 1 and table e-3 (links.lww.com/NXG/ MYORG-related brain calcifications. Apart from copy number A227) present clinical details of all MYORG-related cases in- variants, all types of mutations have been found in this cohort cluded in this study. (1 nonsense, 1 frameshift deletion, 1 insertion, and 9 missense variants). All missense variants were located in conserved and An initial progressive parkinsonism associated with supra- highly conserved amino acid positions (figure 1B). The 12 nuclear gaze palsy phenotype was identified in 1/3 of cases at mutations identified in this study were located throughout the disease onset. Case 1 presented at the initial clinical examina- gene with no obvious mutational hotspots. Two mutations tion age 40 years with profound facial hypomimia with a staring were inherited in the homozygous state in the 2 consanguin- expression, reduced up and down gaze, associated with pro- eous families; the 5 nonconsanguineous families presented with found bilateral bradykinesia, rigidity, reduced arm swings, and compound heterozygous variants. a combination of ataxia and freezing. She had poor response to levodopa. Case 5 presented at age 56 years with asymmetric Phenotype spectrum parkinsonism and supranuclear gaze palsy with poor response In our cohort, MYORG mutation carriers presented with a high to levodopa. Progressive deterioration of motor function, phenotypic variability. The average age at onset was 59.1 years dysarthria, dysphagia, and gait ataxia became evident over the (median 56.4 years, range 39 years to incidental finding at 87 following years. Case 8 presented at age 62 years with par- years). Symptoms at onset varied from parkinsonism (37.5%), kinsonism, frequent falls, staring gaze with vertical gaze palsy ataxia and/or dysarthria (37.5%), and headache (12.5%). that progressed over 16 years along with gait ataxia, cognitive

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 4 erlg:Gntc oue6 ubr2|Arl22 Neurology.org/NG 2020 April | 2 Number 6, Volume | Genetics Neurology:

Table 1 Phenotype description of all MYORG mutations reported in this study

Case number Case 1 Case 2 Case 3 Case 4 Case 5 Case 6 Case 7 Case 8

Family number 12 345 67

Ethnicity/ Caucasian Middle East Caucasian country

cDNA sequence c.176G>A c.1611C>T c.1611C>T c.1634G>A, c.1598C>T c.2211_2212del, c.2162G>A, c.1383C>G c.325C>T, c.1832G>T c.1401_ c.349T>C 1402insCGCTGGTG, c.1967T>C

Amino acid p.Gly59Asp p.Pro496Leu p.Pro496Leu p.Gly545Asp, p.Leu696Profs*10, p.Gly680Ser, p.Pro420Arg p.Gln109Ter, p.Trp426Cysfs*11, change p.Ser533Leu p.Ser117Pro p.Arg611Leu p.Ile656Thr

Zygosity Homozygous Homozygous Compound Compound Compound heterozygous Compound Compound heterozygous heterozygous heterozygous heterozygous

Sex (male/ Female Male Male Female Female Female Male Female female)

Age at 41 52 46 87 72 81 67 68 examination (y)

Age at onset (y) 39 51 45 Incidental finding 56 73 62 62 at 87 y

Disease duration 3 y 1 y 1 y Unknown 16 y 8 y 5 y 6 y

Disability Requires assistance with Normal independent Normal Normal independent Bedridden Requires support for Independent Bedridden and needs most tasks daily living independent daily daily living walking assistance for self-care living

Symptom at Parkinsonism and ataxia Headache Ataxia and Unknown Parkinsonism Progressive dysarthria Progressive dysarthria Parkinsonism onset dizziness and dysphagia

Phenotype

Parkinsonism Yes Yes No Yes Yes Yes Yes Yes

Cerebellar Gait ataxia and Dysarthria Limb ataxia Mild Dysarthria and ataxia Dysarthria Dysarthria Dysarthria syndrome dysarthria dysdiadochokinesia; wide-based gait

Pyramidal Yes No No No Yes No No No syndrome

Dystonia Yes No No No No No No No

Eye Supranuclear gaze palsy Normal Normal Normal Supranuclear gaze No Dysphagia and Supranuclear gaze movements palsy, dysarthria, and dysarthria palsy and dysphagia and cranial dysphagia nerves

Psychiatric No No No No Depression No No Depression symptoms

Continued decline, urinary incontinence, and pyramidal signs. All cases presenting with parkinsonism and supranuclear gaze palsy had associated cognitive impairment characterized by executive dysfunction, poor verbal fluency, and concrete verbal reasoning with low scores on Mini-Mental State Examination (MMSE). No Symmetrical, bilateral basal ganglia, thalami, cerebellar hemispheres, and pons incontinence Two of the 3 patients had a reduced tracer uptake on DAT scan consistent with symmetrical, bilateral marked loss of pre- synaptic dopaminergic neurons (particularly in the putamen).

A cerebellar-bulbar syndrome of variable severity was present in all our cases ranging from very mild (case 4) to moderate dysarthria and dysphagia affecting mainly speech and swal- Cerebellar, pontine and midbrain atrophy Symmetrical, bilateral basal ganglia, thalami, cerebellar hemispheres, and pons lowing (cases 2, 6, and 7). Severe gait and limb ataxia was present in 3/8 of cases (cases 1, 3, and 5). Parkinsonism was detectable in 7 of 8 cases, often associated with other features including supranuclear gaze palsy, early frequent falls, early nt. cognitive decline, and lack of response to levodopa.

One associated extrapyramidal sign in MYORG-related disease atrophy NormalSymmetrical, bilateral basal ganglia, cerebellar hemispheres, and MCI (MOCA 21/30)subcortical white matter MMSE 23/30 was limb dystonia. This was clinically presenting as dystonic posturing in the upper limb precipitated by walking. A third of our patients had bilateral pyramidal signs in the lower limbs. Other associated clinical features were headache (2 cases), urinary in- continence (2 cases), and cramps in the lower limbs (1 case).

Neuropsychiatric evaluation revealed 2 cases with depression. executive dysfunction Symmetrical, bilateral basal ganglia, thalami, cerebellar hemispheres, and pons Cognitive assessment showed impaired cognitive function in 62.5% of cases, with different degrees of severity. MYORG patients showed reduced verbal fluency and poor verbal rea- soning in the first year of disease (cases 1 and 3), mild memory impairment (case 2, MMSE 27/30) with progression over the (continued) following years (case 7, Montreal Cognitive Assessment 21/30 Symmetrical, bilateral basal ganglia, thalami, and cerebellar hemispheres No Urinary incontinence No Noand case 8, MMSE Rhinolalia and urinary 23/30) to a diagnosis of dementia (case 5).

Response to levodopa in cases with parkinsonian phenotype was poor to moderate and proved particularly ineffective in patients with parkinsonism associated with supranuclear gaze palsy. Case 1 with confirmed DAT scan abnormality had some MMSE 28/30 NormalSymmetrical, bilateral basal ganglia and subcortical white matter Dementia and vertigo modest benefit from levodopa in the first year of treatment. However, the response to treatment was short lived and faded in the next 2 years of disease. mutations reported in this study Neuroimaging spectrum All patients showed extensive brain calcifications regardless of MYORG No No No No Bilateral frontotemporal Mild memory impairment MMSE 27/ 30 Symmetrical, bilateral basal ganglia, thalamus, and subcortical white matter disease duration. Basal ganglia (putamen, internal globus pallidus, and caudate nucleus) were involved in all cases, whereas cerebellar hemispheres (folia and dentate nuclei) were involved in 75% of cases. Half of the cases also showed calcification of subcortical white matter. Extensive central pontine calcification was present in 3 cases. Cerebral cortical atrophy was observed in 37% of cases (figure 2A). Bilateral frontotemporal and cerebellar atrophy Reduced verbal fluency, poor Luria, and concrete verbal reasoning Symmetrical, bilateral basal ganglia, cerebellar folia, and subcortical white matter No Headache Headache and

Phenotype description of all Discussion In this study, we screened MYORG mutations in 86 cases from 60 Atrophy Calcification localization Other fi Cognitive function CT results Table 1 Case number Case 1 Case 2 Case 3Abbreviations: cDNA = complementary DNA; MCI = mild cognitive impairment; MMSE = Mini-Mental State Examination; MOCA = Montreal Cognitive Case Assessme 4unrelated, autosomal Case 5 recessive Case 6 PFBC families. We Case 7 identi ed 7 Case 8

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Figure 2 Neuroimaging spectrum in MYORG cases

(A) Neuroimaging spectrum in MYORG cases. Cases 5 (A.a-A.c), 7 (A.d-A.f), and 8 (A.g-A.i) showed similar calcification pattern distribution with extensive involvement of cerebellar dentate nuclei and hemispheres, basal ganglia, thalami, and subcortical white matter; a characteristic central pontine calcification (red arrow) is present in all cases; frontotemporal and cerebellar atrophy was present in case 7; case 6: calcification of the internal globus pallidus, subcortical white matter, and dentate nuclei, with minimal involvement of thalami bilaterally. Severe frontotemporal and cerebellar atrophy is also detectable. (B) MYORG clinical spectrum correlates with MYORG gene expression in different brain areas. MYORG gene expression in different brain areas in adult pathologically normal human brains.25 MYORG is expressed in all 10 brain regions with highest expression detected in the putamen. CRBL = cerebellum; FCTX = frontal cortex; HIPP = hippocampus; MEDU = medulla; OCTX = occipital cortex; PUTM = putamen; SNIG = substantia nigra; TCTX = temporal cortex; THAL = thalamus; WHMT = white matter.

new families of different ethnic backgrounds with disease-causing further extends the phenotypic spectrum of MYORG-related MYORG variants. Biallelic MYORG mutations were associated disease. Of interest, central pontine calcification was present in with PFBCs in 11.6% of families from our cohort. We identified over 1/3 of cases, which seems to be a radiologic diagnostic clue 12 distinct mutations, suggesting that recurrent MYORG muta- for MYORG mutation carriers, as this anatomic region is typi- tions are infrequent. Most of the initial reported cases came from cally not affected in other genetic PFBC cases.6 As physiologic consanguineous families.5,6 Here, we present a cohort largely brain calcifications in this age group are reported in up to lacking in known consanguinity, with the majority of mutations 20%,21,22 an association of calcifications, supranuclear gaze inherited in the compound heterozygous state. palsy, and parkinsonism with atypical features such as ataxia or rapid cognitive decline should prompt physicians to test for Our data suggest that the majority of cases have a disease onset MYORG mutations in this subgroup of patients. We show that in late adulthood with a combination of dysarthria, ataxia, next-generation sequencing can contribute to the diagnosis of parkinsonism, and cognitive decline consistent with the phe- late-onset, mildly affected or asymptomatic cases, therefore – notypes previously reported in MYORG mutations5,6,16 19 and providing a more comprehensive understanding of the genetic other autosomal dominant PFBC-causing genes.20 However, architecture of brain calcifications. parkinsonism with supranuclear gaze palsy was frequently ob- served (37.8% of cases) in our cohort and has not been pre- The exact mechanism leading to disease in MYORG viously described in MYORG mutation carriers. Therefore, this mutations is still unknown. On a cellular level, the gene is

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG expressed in astrocytes localized to the endoplasmic re- Disclosure ticulum5 and playing a role as glycosyl hydrolase.23 Al- The authors report no disclosures. Go to Neurology.org/NG though gene expression (Genotype-Tissue Expression24) for full disclosure. is reported highest in the basal ganglia (nucleus accumbens and caudate) after the skeletal muscle, gene expression 25 Publication history data in BRAINEAC suggest that the putamen and the Received by Neurology: Genetics September 16, 2019. Accepted in final thalamus express the most MYORG messenger RNA fol- form December 17, 2019. lowed by the medulla and the substantia nigra (figure 2B). These areas are mirrored in the clinical phenotype and calcification distribution on neuroimaging assessment in our cohort. Calcifications localized in the basal ganglia Appendix Authors structures (100%), followed by the cerebellum in 75% of Name Location Role Contribution our cases, subcortical white matter (50%), and the thala- Viorica UCL Queen Author Designed and mus (50%). Chelban, MD, Square Institute of conceptualized the MSc, MRCP Neurology, study; acquisition of London, UK data; analyzed the The phenotype observed in individuals with biallelic dele- genetic and clinical terious MYORG variants suggests a high variability among data; and drafted the manuscript for and within families with a disease severity ranging from intellectual content insidious, incidental findings to severe, rapidly progressing Miryam University of Author Acquisition of data; disease course. Asymptomatic cases with biallelic MYORG Carecchio, MD, Padua, Italy interpreted the mutations5 and heterozygous mutation carriers with PhD imaging data; and fi revised the punctate calci cations on the brain CT have been manuscript for 5,6,16 reported ). Our data together with previous reports intellectual suggest a dose-dependent phenotype based on the effect of content mutations on the enzymatic activity of MYORG; however, Gillian Rea, MD Belfast City Author Acquisition of data Hospital, UK and revised the no study has evaluated the enzymatic activity in MYORG manuscript for mutations. intellectual content

Abdalla An-Najah National Author Acquisition of data We show that biallelic MYORG mutations represent a signif- Bowirrat, MD, University, Albiar and revised the icant proportion of PFBC cases without mutations in other PhD Village, Nablus, manuscript for Palestine intellectual content known disease-causing genes. Here, we reported 12 distinct MYORG variants associated with brain calcifications and ex- Salman Aga Khan Author Acquisition of data Kirmani University, and revised the tended the phenotypic spectrum of this disease including Karachi, Pakistan manuscript for atypical parkinsonism with pontine calcification. We recom- intellectual content mend screening MYORG mutations in all patients with pri- Luca Eastern Piedmont Author Acquisition of data mary brain calcifications and autosomal recessive or negative Magistrelli, MD University, and revised the Novara, Italy manuscript for family history. intellectual content

Stephanie UCL Queen Author Acquisition of data Acknowledgment Efthymiou Square Institute of and revised the The authors thank the participants and their families for Neurology, manuscript for London, UK intellectual their essential help with this work. They also acknowledge content C. Panteghini, C. Reale, and F. Invernizzi (Neurogenetics Lucia UCL Queen Author Acquisition of data Unit, Fondazione IRCCS Istituto Neurologico Carlo Besta) Schottlaender, Square Institute of and revised the for their support in sample management. MD, PhD Neurology, manuscript for London, UK intellectual content

Study funding Jana UCL Queen Author Analysis of data and Vandrovcova, Square Institute of revised the This work was supported by The Wellcome Trust (Syn- PhD Neurology, manuscript for aptopathies strategic award [104033]) and the Medical Re- London, UK intellectual content search Council (MRC UK International Centre and project Vincenzo UCL Queen Author Acquisition of data grants). The Association of British Neurologists’ Academic Salpietro, MD Square Institute of and revised the Neurology, manuscript for Clinical Training Research Fellowship andMSA Trust funded London, UK intellectual content V.C. The authors acknowledge the “Cell lines and DNA Bank Ettore Salsano, Fondazione IRCCS Author Acquisition of data of Pediatric Movement Disorders and Mitochondrial Dis- MD Istituto and revised the eases” of the Telethon Network of Genetic Biobanks (grant Neurologico Carlo manuscript for GTB12001J), the EuroBioBank Network, and the Pierfranco Besta, Italy intellectual content and Luisa Mariani Foundation. Continued

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 Appendix (continued) Appendix (continued)

Name Location Role Contribution Name Location Role Contribution

Davide Fondazione IRCCS Author Acquisition of data Henry UCL Queen Author Acquisition of data; Pareyson, MD Istituto and revised the Houlden, PhD Square Institute of revised the Neurologico Carlo manuscript for Neurology, manuscript for Besta, Italy intellectual content London, UK intellectual content; and supervision of Luisa Fondazione IRCCS Author Acquisition of data the entire process Chiapparini, Istituto and revised the MD Neurologico Carlo manuscript for Besta, Italy intellectual content References Farida Jan, MD Aga Khan Author Acquisition of data 1. Nicolas G, Charbonnier C, Campion D, Veltman JA. Estimation of minimal disease University, and revised the prevalence from population genomic data: application to primary familial brain cal- Karachi, Pakistan manuscript for cification. Am J Med Genet B Neuropsychiatr Genet 2018;177:68–74. intellectual content 2. Ramos EM, Carecchio M, Lemos R, et al. Primary brain calcification: an international study reporting novel variants and associated phenotypes. Eur J Hum Genet 2018;26: Shahnaz Aga Khan Author Acquisition of data 1462–1477. Ibrahim University, and revised the 3. Westenberger A, Klein C. The genetics of primary familial brain calcifications. Curr Karachi, Pakistan manuscript for Neurol Neurosci Rep 2014;14:490. intellectual content 4. Taglia I, Bonifati V, Mignarri A, Dotti MT, Federico A. Primary familial brain calci- fication: update on molecular genetics. Neurol Sci 2015;36:787–794. Fatima Khan Aga Khan Author Acquisition of data 5. Yao XP, Cheng X, Wang C, et al. Biallelic mutations in MYORG cause autosomal University, and revised the recessive primary familial brain calcification. Neuron 2018;98:1116–1123.e5. Karachi, Pakistan manuscript for 6. Grangeon L, Wallon D, Charbonnier C, et al. Biallelic MYORG mutation carriers exhibit intellectual content primary brain calcification with a distinct phenotype. Brain 2019;142:1573–1586. 7. Li H, Durbin R. Fast and accurate long-read alignment with Burrows-Wheeler Zul Qarnain Aga Khan Author Acquisition of data transform. Bioinformatics (Oxford, England) 2010;26:589–595. University, and revised the 8. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce Karachi, Pakistan manuscript for framework for analyzing next-generation DNA sequencing data. Genome Res 2010; intellectual content 20:1297–1303. 9. Van der Auwera GA, Carneiro MO, Hartl C, et al. From FastQ data to high confidence Stanislav Institute of Author Acquisition of data variant calls: the Genome Analysis Toolkit best practices pipeline. Curr Protoc Bio- Groppa, PhD Emergency and revised the informatics 2013;43:11.10.1–11.10.33. Medicine, manuscript for 10. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and Chisinau, Republic intellectual content genotyping using next-generation DNA sequencing data. Nat Genet 2011;43: of Moldova 491–498. 11. Danecek P, McCarthy SA. BCFtools/csq: haplotype-aware variant consequences. Nin Bajaj, MD, University of Author Acquisition of data Bioinformatics (Oxford, England) 2017;33:2037–2039. PhD Nottingham, UK and revised the 12. Yang H, Wang K. Genomic variant annotation and prioritization with ANNOVAR manuscript for and wANNOVAR. Nat Protoc 2015;10:1556–1566. intellectual content 13. Auton A, Auton A, Brooks LD, et al. A global reference for human genetic variation. Nature 2015;526:68–74. Bettina Balint, UCL Queen Author Acquisition of data 14. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation MD Square Institute of and revised the in 60,706 humans. Nature 2016;536:285–291. Neurology, manuscript for 15. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of London, UK intellectual content sequence variants: a joint consensus recommendation of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. Genet Kailash P. UCL Queen Author Acquisition of data Med 2015;17:405–424. Bhatia Square Institute of and revised the 16. Arkadir D, Lossos A, Rahat D, et al. MYORG is associated with recessive primary Neurology, manuscript for familial brain calcification. Ann Clin Transl Neurol 2019;6:106–113. London, UK intellectual content 17. Chen Y, Fu F, Chen S, et al. Evaluation of MYORG mutations as a novel cause of primary familial brain calcification. Mov Disord 2019;34:291–297. Andrew Lees, UCL Queen Author Acquisition of data 18. Forouhideh Y, M¨uller K, Ruf W, et al. A biallelic mutation links MYORG to FMedSci, FRCP Square Institute of and revised the autosomal-recessive primary familial brain calcification. Brain 2019;142:e4. Neurology, manuscript for 19. Peng Y, Wang P, Chen Z, Jiang H. A novel mutation in MYORG causes primary London, UK intellectual content familial brain calcification with central neuropathic pain. Clin Genet 2019;95: 433–435. Patrick J. Belfast City Author Acquisition of data 20. Batla A, Tai XY, Schottlaender L, Erro R, Balint B, Bhatia KP. Deconstructing Fahr’s Morrison, CBE, Hospital, UK and revised the disease/syndrome of brain calcification in the era of new genes. Parkinsonism Relat MD, DSc manuscript for Disord 2017;37:1–10. intellectual content 21. K¨onig P. Psychopathological alterations in cases of symmetrical basal ganglia sclerosis. Biol Psychiatry 1989;25:459–468. Nicholas W. UCL Queen Author Acquisition of data 22. Yamada M, Asano T, Okamoto K, et al. High frequency of calcification in basal ganglia Wood, PhD Square Institute of and revised the on brain computed tomography images in Japanese older adults. Geriatr Gerontol Int Neurology, manuscript for 2013;13:706–710. London, UK intellectual content 23. Datta K, Guan T, Gerace L. NET37, a nuclear envelope transmembrane protein with glycosidase homology, is involved in myoblast differentiation. J Biol Chem 2009;284: Barbara Fondazione IRCCS Author Acquisition of data 29666–29676. Garavaglia, Istituto and revised the 24. GTEx Consortium. Human genomics. The genotype-tissue expression (GTEx) pilot PhD Neurologico Carlo manuscript for analysis: multitissue gene regulation in humans. Science 2015;348:648–660. Besta, Italy intellectual content 25. Ramasamy A, Trabzuni D, Guelfi S, et al. Genetic variability in the regulation of gene expression in ten regions of the human brain. Nat Neurosci 2014;17:1418–1428.

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG ARTICLE OPEN ACCESS Heritability of cervical spinal cord structure

Linda Solstrand Dahlberg, PhD, Olivia Viessmann, PhD, and Clas Linnman, PhD Correspondence Dr. Linnman Neurol Genet 2020;6:e401. doi:10.1212/NXG.0000000000000401 [email protected] Abstract Objective Measures of spinal cord structure can be a useful phenotype to track disease severity and development; this observational study measures the hereditability of cervical spinal cord anatomy and its correlates in healthy human beings.

Methods Twin data from the Human Connectome Project were analyzed with semiautomated spinal cord segmentation, evaluating test-retest reliability and broad-sense heritability with an AE model. Relationships between spinal cord metrics, general physical measures, regional brain structural measures, and motor function were assessed.

Results We found that the spinal cord C2 cross-sectional area (CSA), left-right width (LRW), and anterior-posterior width (APW) are highly heritable (85%–91%). All measures were highly correlated with the brain volume, and CSA only was positively correlated with thalamic volumes (p = 0.005) but negatively correlated with the occipital cortex area (p = 0.001). LRW was correlated with the participant’s height (p = 0.00027). The subjects’ sex significantly influenced these metrics. Analyses of a test-retest data set confirmed validity of the approach.

Conclusions This study provides the evidence of genetic influence on spinal cord structure. MRI metrics of cervical spinal cord anatomy are robust and not easily influenced by nonpathological envi- ronmental factors, providing a useful metric for monitoring normal development and pro- gression of neurodegenerative disorders affecting the spinal cord, including—but not limited to—spinal cord injury and MS.

From the Department of Anesthesiology, Perioperative and Pain Medicine (L.S.D., C.L.), Boston Children’s Hospital, Harvard Medical School, MA; Departments of Psychiatry and Radiology (L.S.D., C.L.), Massachusetts General Hospital, Harvard Medical School; Department of Neurology and Neurosurgery (L.S.D.), Montreal Neurological Institute, McGill University, Canada; Athinoula A. Martinos Center for Biomedical Imaging (O.V.), Department of Radiology, Harvard Medical School, Massachusetts General Hospital, Charlestown, Boston; and Spaulding Neuroimaging Lab (C.L.), Spaulding Rehabilitation Hospital, Harvard Medical School, Boston, MA.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary ACE = additive plus common plus nonshared environment and error model; AE = additive plus nonshared environment and error model; AIC = Akaike information criterion; APW = anterior-posterior width; BMI = body mass index; CSA = cross-sectional area; DZ = dizygotic; HCP = Human Connectome Project; LRW = left-right width; MZ = monozygotic; SCI = spinal cord injury.

Automated and semiautomated approaches have been de- In the HCP data set, there is test-retest data from 45 participants. veloped to study spinal cord structure, enabling rater- Of the 45 participants, 9 were excluded because spinal cord independent segmentation and quantification of spinal cord segmentation did not work on either one or both of their scans, metrics. Using these methods, recent studies have reported primarily because of poor tissue contrast or incomplete coverage reductions in the spinal cord cross-sectional area (CSA), left- of the cervical spine. Ultimately, 36 participants remained in the right width (LRW), and anterior-posterior width (APW) in test-retest analysis, where the first data set was also included in MS,1,2 amyotrophic lateral sclerosis,3 and spinal cord injury the main heredity analysis. See Supplemental Data for test-retest – (SCI).4 8 After SCI, changes to the sensorimotor cortex have methodology and results, links.lww.com/NXG/A225. also been reported,9,10 indicative of cortical reorganization because of the lack of afferent input from the spinal cord. Of MRI interest, spinal cord atrophy correlates with physical function- The HCP data were acquired at Washington University in St. ing after SCI.7,8 This suggests that cord atrophy may be pro- Louis on a 3 Tesla Siemens Connectome Skyra scanner (Sie- 13 portional to somato-motor cortex atrophy. However, it is mens, Erlangen, Germany) using a 32-channel head coil. The unknown if such a relationship exists before injury, i.e., is the structural scan was a T1-weighted magnetisation prepared rapid spinal cord structure linked to the cortical sensorimotor rep- gradient echo: repetition time: 2,400 ms, echo time: 2.14 ms, resentation and with motor abilities in healthy subjects? inversion time: 1,000 ms, flip angle: 8°, and matrix size: 266 × 320 × 320, 0.7 mm isotropic voxel size. The 224-mm coverage Human brain anatomy is heritable with a genetic contribution along the Z direction (head to toe) allowed for evaluation of between 66% and 97% for total brain volume, as estimated in cervical structures in most participants to spinal level C2 and in twin studies.11 There are no previous studies on heritability of some cases C3. To take advantage of the full field of view, raw spinal cord structure. Determining factors that contribute to MRI data from HCP were obtained and subsequently corrected variations in spinal cord structure in healthy individuals add to gradient field nonlinearity. See supplemental data for details our understanding of the CNS and, crucially, to markers of (links.lww.com/NXG/A225), including an evaluation of the neurodegenerative pathology. effect of gradient field nonlinearity correction on data validity.

We hypothesized that CSA, LRW, and APW of the spinal cord Spinal cord segmentation Image processing of the spinal cord was carried out with the is (1) reliably measured, (2) hereditary, and (3) is proportional 14 to the volume of the thalamus and cerebellum and the sensory Spinal Cord Toolbox. It used semiautomatic methods for and motor cortex area, as well as to motor function. segmentation, labeling, and extraction of spinal cord metrics. Because the HCP data are centered over the brain, manual landmarks of the spinal cord were used to initiate the detection Methods of the cord for the subsequent automatic segmentation. The output is a binary mask of the spinal cord in 3D space that was Data included in the analyses inspected in each participant. The next step registered the data Data used in the current study were a subset of unprocessed to the MNI-Poly-AMU template, including probabilistic la- structural data from the Human Connectome Project (HCP) beling of the spinal segments of each vertebrae. The template is including test-retest data (db.humanconnectome.org). We then warped back to native space of each participant. The fitof investigated 332 participants. Sufficient brain coverage to the template and each spinal cord segment were manually quantify CSA was obtained in 283 participants. These were 50 inspected in all participants. Finally, CSA, LRW, and APW were pairs of monozygotic (MZ) and 50 pairs of dizygotic (DZ) extracted from each segment of the cord. Here, we examine the twins, as well as 83 unrelated participants. MZ and DZ twin C2 level of the spinal cord because the C2 level is at an ideal pairs were selected to be matched for age (±5 years) and race. location for segmentation and analysis; the surrounding CSF Structural brain scans, behavioral data and information on the creates optimal contrast for accurate segmentation of this area, participants’ whole brain volume (ventricles excluded), and with less curvature than that of the more caudal spinal cord regional brain areas and volumes (obtained by the HCP levels.15 Moreover, studies on SCI and MS have reported the – FreeSurfer parcellation12) were used in the subsequent anal- C2 structure to be linked to clinical outcome scores.16 18 If the yses. Brain regions included the bilateral precentral and border between 2 segments was not in accordance with land- postcentral gyrus, and volumes of the cerebellar gray matter marks surrounding the spinal cord, the slices corresponding to and thalamus. Data for all variables of interest were available each level were manually selected and used in calculation of C2 for all participants. CSA, LRW, and APW.

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Behavioral measurements Heritability analysis All measurements pertaining to motor functioning were To inspect data, correlations in intratwin pairs were first carried obtained with the NIH Toolbox Motor Battery.19 We used out with a Spearman rank-order correlation. A higher correlation data from the 9-hole pegboard dexterity test, the grip in the MZ twins compared with the DZ twins indicates a genetic strength test, the 4-m walk gait speed test, and the 2-minute influence on the tested traits: CSA, LRW, and APW (figure 1). walk endurance test. To estimate simple reaction time, we used reaction times from the 0-back working memory task, Two models were initially run, an additive plus common plus under the assumption that the 0-back control condition nonshared environment and error model (ACE) and an ad- would primarily reflect a direct perceptual response20 not ditive plus nonshared environment and error model (AE) affected by working memory load. model which models the variance in 3 components: additive genetic effect (A), shared environmental effects (C), and Statistical analyses unique environmental effects plus error (E). The model fit Statistical analyses on the test–retest data and spine metrics was estimated with Akaike information criterion (AIC), and behavioral measures were carried out in GraphPad Prism where the lowest AIC value indicates the best fitting model. version 7.0c for Mac (GraphPad Software, La Jolla, California, graphpad.com). Linear regression analyses between cord an- The polygenic model was carried out on CSA, LRW, and APW. atomical metrics, physical, behavioral, and brain metrics were Ageandsexwerethereforeincluded as covariates in the analyses. carried out with the statistical package R in R Studio (RStu- dio: Integrated Development for R. RStudio, Inc., Boston, Standard protocol approvals, registrations, MA, rstudio.com/). Heritability analyses were conducted and patient consents using a classic twin model carried out with the “mets” package, Because this study was conducted with publicly available data also implemented in R. from the HCP (as well as HCP Restricted Access Data), consent was obtained by the HCP. Details regarding data access are found Relations to behavioral and brain measures in their previously published studies (for example [reference 5]). Simple regression models are not appropriate for twin data because the assumption of independence between observations Data availability is violated by the paired structure of data.21 To determine the The imaging data, behavioral test scores, and demographics relationship between spinal cord metrics and behavior or brain used for this project are readily available from the HCP (db. metrics, we used multiple regression models where the mean humanconnectome.org). In accordance with the HCP Re- value of twin pairs and the difference between twin pairs was stricted Access Data Use Terms,22 study-specificparticipant used as regressors (model 2 in reference 21). Data from non- IDs to each individual, as well as the resulting spinal cord twins (n = 83) were also included in the models. segmentation data, will be made available on publication through the HCP Database (db.humanconnectome.org). These analyses were used to investigate the relationship be- tween CSA, LRW, and APW: Results 1. General physical measures: body height, body weight, body mass index (BMI), and total brain volume. Each Demographics model also controlled for the sex of the subjects. The We analyzed C2 CSA, LRW, and APW in 332 participants, resulting 4 separate regression models were Bonferroni whereof 52 participants were excluded from the analyses be- corrected (4 variables, p < 0.0125 considered significant). cause of unreliable or incomplete coverage of the C2 vertebrae. fi 2. Motor function: grip strength (age adjusted), dexterity The nal sample (n = 283) consisted of 190 women and 93 – (age adjusted), endurance (age adjusted) and gait speed, men, with an average age of 29.5 years (range 22 36 years). fi and reaction time from the 0-back working memory task There were signi cantly more women than men in the sample, (5 variables, p < 0.01 considered significant). and women were approximately 2 years younger (women = 3. Brain metrics: area of the bilateral precentral and 27.9 years vs men = 30.1 years, p < 0.0001). Participants in the postcentral gyrus, the volume of the cerebellar gray analyzed sample, consisting mostly of twins and with a higher fi matter, and the volume of the thalamus. As a “control” proportion of females, were on average signi cantly shorter −4 region, we also calculated the relationship to the occipital (168.5 cm vs 171.4 cm, p <10 ) and lighter (75.4 kg vs 79.9 −4 area. Each model also controlled for sex and total brain kg, p <10 ) than the full HCP sample. volume (5 variables, p < 0.01 considered significant). Cord metrics Sex as a biological variable The mean C2 CSA was 71.77 (±5.65) mm2,themeanLRWwas To evaluate if sex influenced CSA, LRW, and APW while 11.52 (±0.61) mm, and the mean APW was 8.0 (±0.49) mm. controlling for weight and body length, 3 multiple regression LRW and APW were highly correlated with CSA values (see models were calculated and adjusted for twin-samples as figure e-1, links.lww.com/NXG/A225); however, LRW and mentioned above. APW were not significantly correlated with each other (table e-1

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Figure 1 Distribution of cervical spinal cord anatomical measures in the sample

(A) CSA = cross-sectional area; (B) A-P = anterior–posterior width; (C) R-L = right–left width.

and figure e-2, links.lww.com/NXG/A225). A robust regression (of which one was also an outlier in the CSA data). We chose not and outlier removal test was carried out to examine the data for to exclude any of these data because a visual inspection verified outliers.23 With the default coefficient criterion of Q = 1%, the test that these were not products of methodological errors, rather they identified 3 outliers in the CSA measures and one outlier in APW represent large values from the natural variability in the data.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 2 C2 cross-sectional T1w image

The segmented cord is marked in blue: (A) largest CSA in sample, 94.2 mm2, (B) smallest CSA in sample, 60.2 mm2, (C) largest ratio between LRW and APW (13.7 mm × 6.6 mm), and (D) smallest ratio between LRW and APW (10.4 mm × 9.1 mm). APW = anterior-posterior width; CSA = cross-sectional area; DZ = dizygotic; LRW = left-right width; MZ = monozygotic.

Figure 1 illustrates C2 anatomical results across the range of The AE model on CSA, LRW, and APW reported a broad- normal variation in the sample, and figure 2 demonstrates the sense heritability of 0.912, 0.852, and 0.868, respectively; see extreme values regarding CSA and ratios between LRW also figure 3 for an illustration. See table 1 for model fitting and APW. parameter estimates and table 2 for heritability estimates.

Heritability analysis Heritability of brain volume (no ventricles) was also carried The AIC of the ACE and AE models for both the CSA APW out. The AIC suggested that the AE model was a better fit for measurements did not differ from each other, suggesting the data. The AE model reported the broad-sense heritability either model describes the data equally well. Comparing the of brain volume to be 0.955. AIC for the LRW measurement resulted in a lower log- likelihood ratio for the AE model, indicating that the AE Correlations with physical, behavioral and model was a better fit for the data, where shared environ- brain measures ment (C) had little influence. This is consistent with pre- We investigated the relationship between CSA, LRW, and vious studies on brain structure.24 APW and (1) general physical measures, (2) motor behavior

Figure 3 Twin pair relationships of spinal cord CSA

Correlations of CSA measures between each (A) MZ and (B) DZ twinset included in the analysis are illustrated, in addition to density plots showing the distribution of the data. CSA = cross-sectional area; DZ = dizygotic; MZ = monozygotic.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Table 1 Model fitting parameter estimates for analysis of heritability of spinal cord metrics and brain volume

Variance estimates

Measure Model 22LL df AIC χ2 p Value a 95% CI c 95% CI e 95% CI

CSA ACE −691.064 5 1,392.13 0.912 0.873–0.951 0.0 0.0 to 0.0 0.088 0.049–0.127

AE −691.064 4 1,390.13 <0.0001 1 0.912 0.873–0.951 0.088 0.049–0.127

APW ACE −124.405 5 258.81 0.868 0.810–0.926 0.0 0.0 to 0.0 0.132 0.074–0.190

AE −124.405 4 256.81 1 0.868 0.810–0.926 0.132 0.074–0.190

LRW ACE −181.56 5 373.112 0.822 0.398–1.246 0.3 −0.389 to 0.149 0.086–0.211 0.448

AE −181.569 4 371.139 0.019 0.8913 0.852 0.789–0.914 0.148 0.086–0.211

Brain ACE −2,993.46 4 5,994.926 0.597 0.369–0.825 0.359 0.128–0.590 0.044 0.025–0.063 volume

AE −2,995.63 3 5,997.255 4.3294 0.0375 0.955 0.936–0.974 0.045 0.026–0.065

Abbreviations: −2LL = −2 log-likelihood; a = additive genetics; ACE = additive plus common plus nonshared environment and error model; AE = additive plus nonshared environment and error model; AIC = Akaike information criterion; APW = anterior-posterior width; c = shared environment; CI = confidence interval; CSA = cross-sectional area; df, degrees of freedom; e = unique environment; LRW = left-right width.

measures, and (3) regional brain measures (see supplementary other hand, CSA was decreased proportional to the area of table e-2, links.lww.com/NXG/A225). Aside from sex, which the occipital cortex (p = 0.001). had a significant influence on several of the models, the only significant coefficients in the general physical models were Sex as a biological variable CSA, LRW, and APW in relation to total brain volume (p =1.7 There was a significant difference in the CSA between men − − − 2 −4 ×10 12,4.2×10 5, and 3.4 × 10 7, respectively) and a signifi- and women (73.57 vs 70.88 mm ; t = 3.86; p <10 ) and in − cance between height and LRW (p = 0.00027), where LRW APW (8.18 vs 7.92 mm; t = 4.43; p <10 4), but not in the increased proportionally with height of the participant. LRW (11.55 vs 11.50 mm; t = 0.62; p = 0.5347).

In the behavioral models, there were no significant rela- A multiple linear regression was calculated to predict CSA tionships. In the regional brain metrics models, whole brain based on sex, body length, and weight, controlling for twin volume was a significant covariate. Significant relationships status A significant regression equation was found (F were observed between CSA and thalamus volume (5,277) = 4.23, p < 0.001), with an R2 of 0.07. Predicted CSA (p = 0.005) and CSA and the occipital area; the CSA in- is equal to 51.08 mm2 − 1.3 (SEX) + 0.3 (HEIGHT) − 0.007 creased with the volume of the thalamus, whereas on the (WEIGHT),wheresexiscodedas0=Male,1=Female,

Table 2 Broad-sense heritability and within-twin correlations for each model of spinal cord and brain volume heritability analyses

Correlation Correlation Measure Model within MZ 95% CI within DZ 95% CI h2

CSA ACE 0.912 0.864–0.944 0.456 0.436–0.475 0.912

AE 0.912 0.864–0.944 0.456 0.436–0.475 0.912

APW ACE 0.868 0.797–0.915 0.434 0.405–0.462 0.868

AE 0.868 0.797–0.915 0.434 0.405–0.462 0.868

LRW ACE 0.852 0.775–0.903 0.441 0.208–0.626 0.822

AE 0.852 0.776–0.903 0.426 0.394–0.457 0.852

Brain volume ACE 0.956 0.933–0.972 0.658 0.523–0.760 0.597

AE 0.955 0.931–0.970 0.477 0.468–0.487 0.955

Abbreviations: a = additive genetics; ACE = additive plus common plus nonshared environment and error model; AE = additive plus nonshared environment and error model; APW = anterior-posterior width; c = shared environment; CI = confidence interval; CSA = cross-sectional area; DZ = dizygotic; e = unique environment; h2 = heritability; LRW = left-right width; MZ = monozygotic.

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG height is measured in inches, and weight is measured in Previous studies have shown large variations in the cervical pounds. Only body length was a predictor of CSA (p = 0.02, CSA of healthy populations, ranging from 70.225 and 79.926 to see table e-1, links.lww.com/NXG/A225). 84.7 mm2.15 Our results are comparable with previous studies using the same segmentation method.27 Postmortem studies A multiple linear regression was calculated to predict LRW have found C2 CSA to be between 56 ± 3.4 mm2,28 70 ± based on sex, body length, and weight, controlling for twin 20 mm2,29 and 83 mm2.30 Data acquisition methods, such status A significant regression equation was found (F (5,277) = using T1- or T2-weighted images31 and data analytical 2.92, p = 0.013), with an R2 of 0.05. Predicted LRW diameter is methods25 influence the measures of cord anatomy. equal to 7.81 mm + 0.2 (SEX) + 0.05 (HEIGHT) − 0.0001 (WEIGHT), where sex is coded as 0 = Male, 1 = Female, height The methods implemented in the Spinal Cord Toolbox have is measured in inches, and weight is measured in pounds. Only been validated elsewhere,31 but here, we had the opportunity body length was a significant predictor of LRW diameter. to test the reproducibility with a test-retest data set of 36 participants. Our analyses resulted in an intraclass correlation A multiple linear regression was calculated to predict APW coefficient measure that is considered good to excellent for based on sex, body length, and weight, controlling for twin CSA and excellent for LRW and APW. Thus, the Spinal Cord status A significant regression equation was found (F (5,277) = Toolbox proves to be a reliable tool for estimating C2 CSA, 4.04, p = 0.001), with an R2 of 0.07. Predicted APW diameter is LRW, and APW from T1-weighted magnetisation prepared equal to 8.13 mm − 0.27 (SEX) + 0.002 (HEIGHT) − 0.0007 rapid gradient echo data, even when C2 is close to the edge of (WEIGHT), where sex is coded as 0 = Male, 1 = Female, height the field of view, if gradient nonlinearity is accounted for. is measured in inches, and weight is measured in pounds. Sex Because our measures displayed high test-retest reliability, the was the only significant predictor of APW diameter. numerical differences in the literature may be due to the dif- ferences in MRI contrast between CSF and white matter dependent on imaging sequences, voxel size, scanner gra- Discussion dients, and differences in segmentation protocols. We rec- ommend caution when pooling data from multiple sites or In a young and healthy cohort, cervical spinal cord structure is multiple imaging sequences. Correction for gradient non- quantifiable using semiautomated and unbiased methods from linearity is crucial (see appendix e-1, links.lww.com/NXG/ brain imaging data. Cervical spinal cord structure is highly A225). Dedicated spinal cord sequences isocentered over the heritable, with genetic influences ranging from 85% to 91%. C2 spine and not the brain, and usage of both T1- and T2- cervical CSA is linearly proportional to total brain volume and weighted images improve generalizability. thalamus volume but is not related to height, weight, BMI, or measures of motor behavior in this sample. Our analyses showed that CSA, LRW, and APW all signifi- cantly correlated with brain volume. Previous studies have Our estimates indicate a genetic component accounting for also reported a relationship between cervical CSA and brain 91% of the variation in spinal cord CSA, 85% for LRW, and volume26,32 or, in postmortem studies, between CSA and 87% for APW. This suggests that the level of genetic influence brain weight.28 We also observed a relationship between LRW on spinal cord structure is comparable with what has been and body height, where LRW increased with height. This is in reported on brain volume (see reference 11 for a review). This line with a previous study reporting a positive relationship suggests that genes play a bigger role in spinal cord structure between C7 CSA and body height in specimens from 152 compared with the nonpathological environment. We also did cadavers.28 However, no such relationship was observed in not observe any relationships to motor behavior in young a study encompassing CT scans of 36 participants.33 healthy controls. As such, the large reductions in CSA con- sistently observed after SCI and in neurodegenerative states We did not observe any further correlations between cord are not likely to be confounded by environmental factors metrics and other physical features or motor behavior. The before the onset of the disease. This makes them useful in only regional brain metric that was correlated with CSA was tracking disease severity and progression. thalamus volume and, surprisingly, an inverse relationship between CSA and occipital area. Given the number of as- The heritability analyses showed that shared environmental cending spinal cord tracts projecting to the thalamus, this influence (the C component in the ACE model) had close to relationship is plausible. The negative relationship between no influence in the 3 different measurements. The lack of cord CSA and occipital area is more surprising and may be influence by a shared environment could be because of the because of multicollinearity between occipital area and total assumption that both DZ and MZ twins share a more similar brain volume. environment, both in utero and in childhood, compared with nontwin family members. As in previous studies,26,34 men had significantly higher CSA than women and also higher LRW and APW. Notably, in The HCP sample showed an average C2 CSA of 71.77 mm2 multiple regression models, body length was the only pre- (n = 283), LRW of 11.52 mm, and an APW of 8.0 mm. dictor of CSA and LRW, whereas sex was the only predictor

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 of APW. It has been suggested that LRW is reflective of provide additional meaningful metrics both for clinical and motor tracts mainly located in the lateral funiculi, whereas scientific examination. APW is reflective of the sensory tracts found in the dorsal funiculus.35 Indeed, we found that LRW and APW were not Some caution should be exercised when interpreting the proportional and thus largely independent metrics of cord spinal cord imaging studies because the spinal cord is a rela- anatomy. Previous studies indicate that tactile spatial acuity tively small area and is susceptible to partial volume effects. improves with decreasing finger size independent of sex36 HCP data were collected using 0.7 × 0.7 × 0.7 mm resolution, and that fingertips have a similar number of Meissner cor- a substantial improvement over typical 1-mm isotropic data puscles, regardless of size.37 As such, the number of fingertip but much higher resolution methods are being developed.49 sensory axons at C2 would be similar for a small and a large Moreover, signal-to-noise ratio at the outer edges of the field hand (or, in effect, for a short and a tall person). If this of view (i.e., in the spine area of a brain scan) can be low, extrapolation holds for the whole body, it would suggest that making a segmentation based on intensities more challenging. the total number of sensory receptors and associated spinal However, the C2 spinal cord level is an optimal region to axons are roughly equivalent across men and women and study because there is very little curvature, making a distinc- across body size but with higher density in smaller bodies. tion between the spinal cord and surrounding CSF easier.15 The observed relation between body length and cord di- ameter would then be reflective of the average axonal di- Another limitation on the analytic level is the use of the AE ameter, rather than the number of axons. Because axonal model. The AE model gives estimation pertaining to 2 factors: conduction velocity in myelinated axons is approximately additive genetic effect (A) and unique environmental effect linearly proportional to axon diameter,38 we speculate that (E). It is noted that the E term also absorbs variation that the observed cord-thickness body-length relationship is re- arises from measurement error and individual day-to-day flective of increased axonal diameter to achieve similar fluctuations. Linear mixed effects models that explicitly ac- transmission times in short and tall bodies. count measurement errors by using repeated measures have been developed50 but were not used here because our test- Owing to the narrow age range in the HCP young adult retest sample was deemed too small. sample, we did not evaluate age effects. Previous studies are mixed, with reports of no correlation with age, height, and Similar to the brain, cervical spinal cord anatomy is highly weight,33 as well as reports of a relationship between spinal heritable. Provided that the field of view is sufficient to cover cord CSA and age and height.39 the first 2–3 vertebrae, C2 CSA, LRW, and APW can reliably be measured in brain dedicated neuroimaging protocols. With Previous studies in the spinal cord injured population have large data sharing initiatives, this opens the possibility to ex- demonstrated parallel changes in both cord CSA and so- amine these relatively unexplored metrics that harbor im- matosensory regions7,40,41 between CSA and hand grip portant markers of development and pathology. strengths,18 and between LRW and motor score,35 whereas APW correlated with sensory scores.35 In patients with MS, Acknowledgment atrophy of the upper cervical cord is evident in APW but not Data were provided by the Human Connectome Project, LRW,42 whereas studies on ALS have only reported on CSA.3 WU-Minn Consortium (Principal Investigators: David Van We did not observe any relationships between motor function Essen and Kamil Ugurbil; 1U54MH091657) funded by the 16 and cord metrics in the present large, young, and healthy NIH Institutes and Centers that support the NIH Blueprint cohort. This suggests that it might only be in pathologic states for Neuroscience Research and by the McDonnell Center for with anterograde and retrograde degeneration of white mat- Systems Neuroscience at Washington University. ter, reducing cord area by 5–22 mm2, that such relationships are unmasked. Study funding Research reported in this publication was supported by the Several large imaging studies in MS have demonstrated ex- Promobilia Foundation, Wings for Life, The Gordon Foun- – tensive cord atrophy.42 44 We also know from longitudinal dation (C. Linnman), the Department of Defense SC140194 neuroimaging studies that brain volume decreases with aging (C. Linnman, L. Solstrand Dahlberg), NICHD of the NIH and in neurodegenerative disorders such as Alzheimer and under award number 1R01HD097407 (C. Linnman), and Parkinson disease. Whether the CSA of the spinal cord changes the BRAIN Initiative (NIMH grant R01-MH111419) (O. over time in healthy individuals is inconclusive.45 Some studies Viessmann). have found small reductions in the spinal cord CSA in elderly – individuals.46 48 Future studies should aim to elucidate the Disclosure changes in spinal cord structure in healthy and pathologic aging Disclosures available: Neurology.org/NG. and if it correlates with changes in motor and sensory func- tioning. Several ongoing brain neuroimaging efforts have an Publication history adequate field of view to evaluate developmental and neuro- Received by Neurology: Genetics May 17, 2019. Accepted in final form degenerative effects on the upper cervical cord. This will January 13, 2020.

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 22. Van Essen DC, Smith SM, Barch DM, Behrens TEJ, Yacoub E, Ugurbil K. The WU- – Appendix 1 Authors Minn Human Connectome Project: an overview. Neuroimage 2013;80:62 79. 23. Motulsky HJ, Brown RE. Detecting outliers when fitting data with nonlinear re- gression - a new method based on robust nonlinear regression and the false discovery Name Location Role Contribution rate. BMC Bioinformatics 2006;7:123. 24. Hibar DP, Stein JL, Renteria ME, et al. Common genetic variants influence human Linda Harvard Author Analyzed the data, data subcortical brain structures. Nature 2015;520:224–229. Solstrand Medical interpretation, and drafted the 25. Yiannakas MC, Mustafa AM, De Leener B, et al. Fully automated segmentation of the Dahlberg, School, MA manuscript cervical cord from T1-weighted MRI using PropSeg: application to multiple sclerosis. PhD Neuroimage Clin 2016;10:71–77. 26. Papinutto N, Schlaeger R, Panara V, et al. Age, gender and normalization covariates Olivia Harvard Author Processed the data and for spinal cord gray matter and total cross-sectional areas at cervical and thoracic Viessmann, Medical revised the manuscript for levels: a 2D phase sensitive inversion recovery imaging study. PLoS One 2015;10: PhD School, MA intellectual content e0118576. 27. De Leener B, Cohen-Adad J, Kadoury S. Automatic segmentation of the spinal cord Clas Harvard Author Conceptualized the study, and spinal canal coupled with vertebral labeling. IEEE Trans Med Imaging 2015;34: Linnman, Medical analyzed the data, data 1705–1718. PhD School, MA interpretation, and drafted the 28. Kameyama T, Hashizume Y, Ando T, Takahashi A. Morphometry of the normal manuscript cadaveric cervical spinal cord. Spine (Phila Pa 1976) 1994;19:2077–2081. 29. Ko H-Y, Park JH, Shin YB, Baek SY. Gross quantitative measurements of spinal cord segments in human. Spinal Cord 2004;42:35–40. 30. Donaldson HH, Davis DJ. A description of charts showing the areas of the cross References sections of the human spinal cord at the level of each spinal nerve. J Comp Neurol 1. Rashid W, Davies GR, Chard DT, et al. Increasing cord atrophy in early relapsing- 1903;13:19–40. remitting multiple sclerosis: a 3 year study. J Neurol Neurosurg Psychiatry 2006;77: 31. De Leener B, Kadoury S, Cohen-Adad J. Robust, accurate and fast automatic seg- 51–55. mentation of the spinal cord. Neuroimage 2014;98:528–536. 2. Abdel-Aziz K, Schneider T, Solanky BS, et al. Evidence for early neurodegeneration in 32. Engl C, Schmidt P, Arsic M, et al. Brain size and white matter content of cerebrospinal the cervical cord of patients with primary progressive multiple sclerosis. Brain 2015; tracts determine the upper cervical cord area: evidence from structural brain MRI. 138:1568–1582. Neuroradiology 2013;55:963–970. 3. El Mendili M-M, Cohen-Adad J, Pelegrini-Issac M, et al. Multi-parametric spinal cord 33. Yu YL, du Boulay GH, Stevens JM, Kendall BE. Morphology and measurements of the MRI as potential progression marker in amyotrophic lateral sclerosis. PLoS One cervical spinal cord in computer-assisted myelography. Neuroradiology 1985;27: 2014;9:e95516. 399–402. 4. Filippi M, Rocca MA. Multiple sclerosis: linking disability and spinal cord imaging 34. Ulbrich EJ, Schraner C, Boesch C, et al. Normative MR cervical spinal canal outcomes in MS. Nat Rev Neurol 2013;9:189–190. dimensions. Radiology 2014;271:172–182. 5. Kearney H, Miller DH, Ciccarelli O. Spinal cord MRI in multiple sclerosis— 35. Lundell H, Barthelemy D, Skimminge A, Dyrby TB, Biering-Sorensen F, Nielsen JB. diagnostic, prognostic and clinical value. Nat Rev Neurol 2015;11:327–338. Independent spinal cord atrophy measures correlate to motor and sensory deficits in 6. Lin X, Tench CR, Evangelou N, Jaspan T, Constantinescu CS. Measurement of spinal individuals with spinal cord injury. Spinal Cord 2011;49:70–75. cord atrophy in multiple sclerosis. J Neuroimaging 2004;14:20S–26S. 36. Peters RM, Hackeman E, Goldreich D. Diminutive digits discern delicate details: 7. Freund P, Weiskopf N, Ashburner J, et al. MRI investigation of the sensorimotor fingertip size and the sex difference in tactile spatial acuity. J Neurosci 2009;29: cortex and the corticospinal tract after acute spinal cord injury: a prospective longi- 15756–15761. tudinal study. Lancet Neurol 2013;12:873–881. 37. Dillon YK, Haynes J, Henneberg M. The relationship of the number of Meissner’s 8. Freund P, Curt A, Friston K, Thompson A. Tracking changes following spinal cord corpuscles to dermatoglyphic characters and finger size. J Anat 2001;199:577–584. injury: insights from neuroimaging. Neuroscientist 2013;19:116–128. 38. Hursh JB. Conduction velocity and diameter of nerve fibers. Am J Physiol Content 9. Jurkiewicz MT, Crawley AP, Verrier MC, Fehlings MG, Mikulis DJ. Somatosensory 1939;127:131–139. cortical atrophy after spinal cord injury: a voxel-based morphometry study. Neurology 39. Kameyama T, Hashizume Y, Sobue G. Morphologic features of the normal human 2006;66:762–764. cadaveric spinal cord. Spine (Phila Pa 1976) 1996;21:1285–1290. 10. Wrigley PJ, Gustin SM, Macey PM, et al. Anatomical changes in human motor cortex 40. Jutzeler CR, Huber E, Callaghan MF, et al. Association of pain and CNS structural and motor pathways following complete thoracic spinal cord injury. Cereb Cortex changes after spinal cord injury. Sci Rep 2016;6:18534. 2009;19:224–232. 41. Hou J, Xiang Z, Yan R, et al. Motor recovery at 6 months after admission is related to 11. Peper JS, Brouwer RM, Boomsma DI, Kahn RS, Hulshoff Pol HE. Genetic influences structural and functional reorganization of the spine and brain in patients with spinal on human brain structure: a review of brain imaging studies in twins. Hum Brain cord injury. Hum Brain Mapp 2016;37:2195–2209. Mapp 2007;28:464–473. 42. Lundell H, Svolgaard O, Dogonowski AM, et al. Spinal cord atrophy in anterior- 12. Glasser MF, Coalson TS, Robinson EC, et al. A multi-modal parcellation of human posterior direction reflects impairment in multiple sclerosis. Acta Neurol Scand 2017; cerebral cortex. Nature 2016;536:171–178. 136:330–337. 13. Glasser MF, Sotiropoulos SN, Wilson JA, et al. The minimal preprocessing pipelines 43. Azodi S, Nair G, Enose-Akahata Y, et al. Imaging spinal cord atrophy in progressive for the Human Connectome Project. Neuroimage 2013;80:105–124. myelopathies: HTLV-I-associated neurological disease (HAM/TSP) and multiple 14. De Leener B, Levy S, Dupont SM, et al. SCT: Spinal Cord Toolbox, an open-source sclerosis (MS). Ann Neurol 2017;82:719–728. software for processing spinal cord MRI data. Neuroimage 2017;145:24–43. 44. Eden D, Gros C, Badji A, et al. Spatial distribution of multiple sclerosis lesions in the 15. Losseff NA, Webb SL, O’Riordan JI, et al. Spinal cord atrophy and disability in cervical spinal cord. Brain 2019;142:633–646. multiple sclerosis: a new reproducible and sensitive MRI method with potential to 45. Valsasina P, Rocca MA, Horsfield MA, Copetti M, Filippi M. A longitudinal MRI monitor disease progression. Brain 1996;119:701–708. study of cervical cord atrophy in multiple sclerosis. J Neurol 2015;262:1622–1628. 16. Ziegler G, Grabher P, Thompson A, et al. Progressive neurodegeneration following 46. Valsasina P, Horsfield MA, Rocca MA, Absinta M, Comi G, Filippi M. Spatial nor- spinal cord injury: implications for clinical trials. Neurology 2018;90:e1257–e1266. malization and regional assessment of cord atrophy: voxel-based analysis of cervical 17. Horsfield MA, Sala S, Neema M, et al. Rapid semi-automatic segmentation of the cord 3D T1-weighted images. AJNR Am J Neuroradiol 2012;33:2195–2200. spinal cord from magnetic resonance images: application in multiple sclerosis. Neu- 47. Agosta F, Lagan`a M, Valsasina P, et al. Evidence for cervical cord tissue dis- roimage 2010;50:446–455. organisation with aging by diffusion tensor MRI. Neuroimage 2007;36:728–735. 18. Freund P, Weiskopf N, Ward NS, et al. Disability, atrophy and cortical reorganization 48. Yanase M, Matsuyama Y, Hirose K, et al. Measurement of the cervical spinal cord following spinal cord injury. Brain 2011;134:1610–1622. volume on MRI. J Spinal Disord Tech 2006;19:125–129. 19. Reuben DB, Magasi S, McCreath HE, et al. Motor assessment using the NIH 49. Zhao W, Cohen-Adad J, Polimeni JR, et al. Nineteen-channel receive array and four- Toolbox. Neurology 2013;80:S65–S75. channel transmit array coil for cervical spinal cord imaging at 7T. Magn Reson Med 20. Hur J, Iordan AD, Dolcos F, Berenbaum H. Emotional influences on perception and 2014;72:291–300. working memory. Cogn Emot 2017;31:1294–1302. 50. Ge T, Holmes AJ, Buckner RL, Smoller JW, Sabuncu MR. Heritability analysis with 21. Carlin JB, Gurrin LC, Sterne JA, Morley R, Dwyer T. Regression models for twin repeat measurements and its application to resting-state functional connectivity. Proc studies: a critical review. Int J Epidemiol 2005;34:1089–1099. Natl Acad Sci USA 2017;114:5521–5526.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 ARTICLE OPEN ACCESS Mitochondrial diseases in North America An analysis of the NAMDC Registry

Emanuele Barca, MD, PhD, Yuelin Long, MS, Victoria Cooley, MS, Robert Schoenaker, MD, BS, Correspondence Valentina Emmanuele, MD, PhD, Salvatore DiMauro, MD, Bruce H. Cohen, MD, Amel Karaa, MD, Dr. Hirano [email protected] Georgirene D. Vladutiu, PhD, Richard Haas, MBBChir, Johan L.K. Van Hove, MD, PhD, Fernando Scaglia, MD, Sumit Parikh, MD, Jirair K. Bedoyan, MD, PhD, Susanne D. DeBrosse, MD, Ralitza H. Gavrilova, MD, Russell P. Saneto, DO, PhD, Gregory M. Enns, MBChB, Peter W. Stacpoole, MD, PhD, Jaya Ganesh, MD, Austin Larson, MD, Zarazuela Zolkipli-Cunningham, MD, Marni J. Falk, MD, Amy C. Goldstein, MD, Mark Tarnopolsky, MD, PhD, Andrea Gropman, MD, Kathryn Camp, MS, RD, Danuta Krotoski, PhD, Kristin Engelstad, MS, Xiomara Q. Rosales, MD, Joshua Kriger, MS, Johnston Grier, MS, Richard Buchsbaum, John L.P. Thompson, PhD, and Michio Hirano, MD

Neurol Genet 2020;6:e402. doi:10.1212/NXG.0000000000000402 Abstract Objective To describe clinical, biochemical, and genetic features of participants with mitochondrial diseases (MtDs) enrolled in the North American Mitochondrial Disease Consortium (NAMDC) Registry.

Methods This cross-sectional, multicenter, retrospective database analysis evaluates the phenotypic and molecular characteristics of participants enrolled in the NAMDC Registry from September 2011 to December 2018. The NAMDC is a network of 17 centers with expertise in MtDs and includes both adult and pediatric specialists.

Results One thousand four hundred ten of 1,553 participants had sufficient clinical data for analysis. For this study, we included only participants with molecular genetic diagnoses (n = 666). Age at onset ranged from infancy to adulthood. The most common diagnosis was multisystemic disorder (113 participants), and only a minority of participants were diagnosed with a classical mitochondrial syndrome. The most frequent classical syndromes were Leigh syndrome (97 individuals) and mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes (71 individuals). Pathogenic variants in the mitochondrial DNA were more frequently observed (414 participants) than pathogenic nuclear gene variants (252 participants). Pathogenic variants in 65 nuclear genes were identified, with POLG1 and PDHA1 being the most commonly affected. Pathogenic variants in 38 genes were reported only in single participants.

Conclusions The NAMDC Registry data confirm the high variability of clinical, biochemical, and genetic features of participants with MtDs. This study serves as an important resource for future enhancement of MtD research and clinical care by providing the first comprehensive description of participant with MtD in North America.

From the Department of Neurology (E.B., V.E., S.D., K.E., X.Q.R., M.H.), Columbia University Medical Center, New York; Department of Biostatistics (Y.L., V.C., J.K., J. Grier, R.B., J.L.P.T.), Mailman School of Public Health, Columbia University, New York; Radboudumc (R.S.), Nijmegen, The Netherlands; Department of Pediatrics (B.H.C.), Northeast Ohio Medical University and Akron Children’s Hospital; Genetics Unit (A.K.), Massachusetts General Hospital, Boston; Department of Pediatrics (G.D.V.), State University of New York at Buffalo; Departments of Neurosciences and Pediatrics (R.H.), University of California at San Diego; Department of Pediatrics (J.L.K.V.H., A.L.), University of Colorado School of Medicine, Aurora; Department of Molecular and Human Genetics (F.S.), Baylor College of Medicine, Houston, TX; Texas Children’s Hospital (F.S.), Houston; Joint BCM-CUHK Center of Medical Genetics (F.S.), Prince of Wales Hospital, ShaTin, New Territories, Hong Kong; Department of Neurology (S.P.), Cleveland Clinic, OH; Departments of Genetics and Genome Sciences and Pediatrics (J.K.B., S.D.D.), and Center for Human Genetics, University Hospitals Cleveland Medical Center, Case Western Reserve University, OH; Departments of Neurology and Clinical Genomics (R.H.G.), Mayo Clinic, Rochester, MN; Department of Neurology (R.P.S.), University of Washington, Seattle Children’s Hospital; Department of Pediatrics (G.M.E.), Stanford University, Palo Alto, CA; Department of Medicine (P.W.S.), University of Florida at Gainesville; Genetics and Genomic Sciences at the Icahn School of Medicine at Mount Sinai (J. Ganesh), New York; Mitochondrial Medicine Frontier Program (Z.Z.-C., M.J.F., A.C.G.), Division of Human Genetics, The Children’s Hospital of Philadelphia and University of Pennsylvania Perelman School of Medicine; University of Pennsylvania Perelman School of Medicine (Z.Z.-C.), Philadelphia; Department of Neurology (M.T.), McMasters University, Toronto, Ontario, Canada; Department of Neurology (A.G.), Children’s National Health Network, Washington, DC; Office of Dietary Supplements (K.C.), National Institutes of Health, Bethesda, MD; and Eunice Kennedy Shriver National Institute of Child Health and Human Development (D.K.), National Institutes of Health, Bethesda, MD.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by NIH U54 NS078059. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary

CoQ10 = coenzyme Q10; COX = cytochrome c oxidase; cPEO = chronic progressive external ophthalmoplegia; IRB = institutional review board; LS = Leigh syndrome; LHON = Leber hereditary optic neuropathy; MELAS = mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes; MERRF = myoclonus epilepsy with ragged red fibers; MtD = mitochondrial disease; mtDNA = mitochondrial DNA; NAMDC = North American Mitochondrial Disease Consortium; nDNA = nuclear DNA; OxPhos = oxidative phosphorylation; PDC = pyruvate dehydrogenase complex; SLE = stroke-like episode; TP = thymidine phosphorylase.

Mitochondria are the organelles that generate cellular energy history, physical examination, and laboratory tests and pro- (adenosine triphosphate) via oxidative phosphorylation vide written consent using IRB-approved forms. Clinicians (OxPhos). OxPhos is an elaborate multiprotein machine input data using a secure web-based data entry system. Data composed of a series of enzyme complexes embedded in the were collected through December 1, 2018. inner mitochondrial membrane (complex I–V).1 Because virtually all tissues require mitochondria to function, mito- Data availability chondrial dysfunction manifests commonly as multisystem The NAMDC Clinical Registry data are stored in a secure disorders, with frequent involvement of high-energy demand database and are available to other investigators on sub- tissues, such as brain, muscle, and heart.2 The OxPhos system mission of a data use application and approval by the consists of 85 subunits and is under the control of 2 genomes, NAMDC Data Use Committee. the nuclear DNA (nDNA) and mitochondrial DNA (mtDNA). Mitochondrial diseases (MtDs) can arise due to Clinical diagnosis pathogenic variants in either of the 2 genomes.3 Because of Participants are enrolled in the Registry based on the ’ the genetic complexity and the multiple biochemical functions NAMDC site investigators clinical diagnoses of MtDs. Di- of these organelles, MtDs are phenotypically and genetically agnoses are based on established and published clinical 6,7 fi ff heterogeneous. Because of this vast diversity, these disorders criteria. Twenty- ve di erent clinical diagnoses were are challenging to diagnose, manage clinically, and investigate coded: 15 classical mitochondrial clinical syndromes and 8 fi with an association of symptoms and signs commonly ob- scienti cally. Epidemiologically, MtDs are considered rare 8–12 diseases, but among rare disorders, they are relatively fre- served in mitochondrial participants (table 1). Some of quent, with an overall estimated prevalence of 11.5:100,000.3 the syndromes require an association of clinical and radiologic fi In totality, MtDs represent the most prevalent group of ndings, e.g., for the diagnosis of leukoencephalopathy, white inherited neurologic disorders.4 Disease registries constitute matter lesions evident on brain MRI are typically associated a cornerstone of MtDs patient care because they help to es- with cognitive impairment, long-tract signs, or both. En- fi tablish a reliable picture of the distribution and characteristics cephalopathy was de ned by the presence of dementia, seiz- of the participants, focus resources, and gather accurate data ures, corticospinal tract dysfunction, movement disorders, or for efficient clinical mechanistic studies and therapeutic trials.5 combinations of these manifestations due to cerebral This article reviews the spectrum of clinical, biochemical, and pathology. molecular genetic features of MtD participants enrolled in the fi North American Mitochondrial Disease Consortium Two nonspeci c groups of participants with complex clinical (NAMDC) Registry (hereafter referred to as the Registry). manifestations that do not fall in any of the other categories are (1) multisystemic and (2) other clinical disorders. For multisystemic disease, clinical manifestations must involve at Methods least 3 organs with a progressive clinical course that does not otherwise conform to a classical phenotype. Participants Standard protocol approvals, registrations, without clinical manifestations were classified as asymptom- and patient consents atic carriers, which are not included in this analysis. The NAMDC is an NIH-funded collaborative effort of 17 North American medical centers, which work in close part- Clinical, biochemical, and laboratory data nership with the United Mitochondrial Disease Foundation Clinical, laboratory, and biochemical data were collected: fi and other patient advocacy groups. Since 2011, the NAMDC de ciencies of OxPhos activities, coenzyme Q10 (CoQ10), has maintained a Registry that has collected clinical, bio- pyruvate dehydrogenase complex (PDC), or thymidine chemical, histologic, and molecular genetic data on partic- phosphorylase (TP). ipants with MtDs. The Registry protocol was approved by the NIH and the NAMDC central Institutional Review Board Molecular genetics (IRB) with local context review at each NAMDC site nDNA pathogenic variants as well as mtDNA pathogenic point (NCT01694940). To enroll, participants had to visit an ex- variants, single and multiple large-scale deletions, and depletion perienced clinician at one of the sites for a review of medical were recorded. All genetic variants in the database were also

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 1 Demographic, clinical, biochemical, and genetic Table 1 Demographic, clinical, biochemical, and genetic characteristics of the North American characteristics of the North American Mitochondrial Disease Consortium Clinical Mitochondrial Disease Consortium Clinical Registry participants (N = 666) Registry participants (N = 666) (continued)

Characteristic (participants) Frequency (%) Characteristic (participants) Frequency (%)

Sex NARPa 8 (1.2)

Female 380 (57.1) Hepatocerebral syndrome 6 (0.9)

Male 285 (42.8) Maternal inherited deafness 5 (0.8)

Missing 1 (0.2) Cardiomyopathy 5 (0.8)

Racial composition Leukoencephalopathy 4 (0.6)

White 567 (85.1) RIM with COX deficiencya 2 (0.3)

Asian 24 (3.6) Barth syndromea 2 (0.3)

More than one 22 (3.3) Biochemical deficiencies (114 recorded; not recorded in 552 [82.9%]) Black/African American 19 (2.9) Isolated complex I 18 (2.7) Other 33 (5.0) Isolated complex II 1 (0.2) Missing 1 (0.2) Isolated complex III 1 (0.2) Age at onset Isolated complex IV 14 (2.1) <2 y 198 (29.7) Isolated complex V 5 (0.8) ≥2to<5y 83 (12.5) Combined RCE complexes 21 (3.2) ≥5 to <12 y 76 (11.4)

CoQ10 2 (0.3) ≥12 to <18 y 52 (7.8) TP 8 (1.2) ≥18 y 180 (27.0) PDC 44 (6.6) Missing 77 (11.6) Molecular genetic information Clinical syndrome mtDNA pathogenic variants 414 (62.2) Multisystemic syndrome 113 (17.0) nDNA pathogenic variants 252 (37.8) LSa 97 (14.6) Abbreviations: CoQ = coenzyme Q ; cPEO = chronic progressive external Other clinical syndrome 75 (11.3) 10 10 ophthalmoplegia; KSS = Kearns-Sayre syndrome; LHON = Leber hereditary

a optic neuropathy; LS = Leigh syndrome; MELAS = mitochondrial encepha- MELAS 71 (10.7) lomyopathy, lactic acidosis, and stroke-like episodes; MERRF = myoclonus epilepsy with ragged red fibers; MIDD = maternal inherited diabetes and Encephalopathy 38 (5.7) deafness, also known as diabetes and deafness, DAD; MNGIE = mitochon- drial neurogastrointestinal encephalopathy; mtDNA = mitochondrial DNA; cPEO+a 37 (5.6) NARP = neuropathy, ataxia, and retinitis pigmentosa; nDNA = nuclear DNA; PDC = pyruvate dehydrogenase complex; RCE = respiratory chain enzyme; Encephalomyopathy 32 (4.8) RIM with COX deficiency = reversible infantile myopathy with cytochrome c oxidase deficiency; SANDO = sensory ataxic neuropathy with dysarthria and LHONa 28 (4.2) ophthalmoparesis; TP = thymidine phosphorylase. Multisystemic syndrome is defined by clinical manifestations involving at least 3 organs. Other clinical syndrome is defined as a disorder not con- Myopathy 25 (3.8) forming to one of the classical syndromes or multisystemic syndrome. a Classical syndromes. SANDOa 19 (2.9)

cPEOa 18 (2.7)

KSSa 17 (2.6) assessed for pathogenicity using MseqDR (mseqdr.org), and variants in the POLG1 gene were assessed using the NIH-based a MIDD 16 (2.4) POLG1 mutation database (tools.niehs.nih.gov/polg/). Alpers syndromea 15 (2.3) Statistical analysis a MERRF 13 (2.0) The Fisher exact test (SAS 9.4: SAS Institute Inc., Cary, NC) Pearson syndromea 10 (1.5) was used in the comparisons of participants with different genetic features, and Holm adjustment (R 3.5.1: Bell Labo- MNGIEa 10 (1.5) ratories, New Providence, NJ) for multiple comparisons.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Results adulthood (table 2). Apart from key clinical findings in CNS and skeletal muscle involvement, heart abnormalities were Of the 1,533 total participants enrolled in the Registry (Jan- recorded (21 participants), with cardiac arrhythmias being the uary 2011–December 2018), 1,410 had complete data on most prevalent (table 2, table e-1, links.lww.com/NXG/ clinical manifestations, biochemistry, and genetics. Genetic A226). Gastrointestinal tract manifestations and diabetes diagnosis was available for 722 participants. After the exclu- mellitus were also commonly reported (table 2, table e-2). In sion of 56 asymptomatic carriers, 666 genetically diagnosed addition to stroke-like episodes (SLEs), the most common symptomatic individuals remained for analysis (table 1). CNS manifestations were seizures, migraine, and ataxia (tables e-1 and e-2). Thirty-nine participants had constitu- Demographic characteristics are summarized in table 1. Age at tional symptoms (short stature and low body mass index) disease onset showed a bimodal distribution with peaks below (table 2). Genetically, m.3243A>G in the mitochondrial age 2 years and in adulthood (≥18 years). The most frequent MTTL1 gene was the most frequently observed pathogenic classical diagnoses were Leigh syndrome (LS) (97 partic- variant. In addition, 10 of 69 participants (14.5%) with POLG ipants), mitochondrial encephalomyopathy, lactic acidosis, pathogenic variants manifested SLEs, but only 1 was di- and stroke-like episodes (MELAS) (71 participants), chronic agnosed with MELAS. progressive external ophthalmoplegia (cPEO) and cPEO with other organ-system manifestations (beside skeletal muscle Chronic progressive external ophthalmoplegia involvement) (cPEO+) (55 participants), and Leber heredi- Fifty-five participants were diagnosed with cPEO/cPEO+ (18 tary optic neuropathy (LHON) (28 participants) (table 1). A cPEO and 37 cPEO+). The majority were adults (table 2). large proportion of participants had nonspecific disease syn- Muscle involvement in cPEO was not confined to the ex- dromes (298 participants). These nonspecific disorders in- trinsic ocular muscles, as skeletal myopathy (10 participants) cluded multisystemic (113 participants) and other clinical and dysphagia (4 participants) were also observed. Partic- syndromes (75 participants) (table 1). The prevalence of ipants with cPEO+ frequently had CNS symptoms (24 par- nonspecific diagnoses was high in both the pediatric (<18 ticipants), with ataxia and hearing loss being the most years) and adult population (≥18 years). prevalent (table e-1, links.lww.com/NXG/A226). Both mtDNA and nDNA pathogenic variants have been observed Classical mitochondrial syndromes (figure 1). Variants in mtDNA were more frequent than Leigh syndrome nDNA in both groups (14 cPEO participants and 23 cPEO+ Ninety-seven participants had LS. Disease onset was seen participants). Single large-scale mtDNA deletions were predominantly in infancy (<2 years in 55 participants) (table present in 12 participants with cPEO and 15 with cPEO+. fi 2). The most frequent manifestations were developmental Although infrequently identi ed, nDNA pathogenic variants delay (84 participants), followed by developmental regression were more common in the cPEO+ (11 participants) group; (42 participants). Other neurologic manifestations included the majority had variants in POLG1 (6 cPEO+ and 2 cPEO). ataxia (44 participants), dystonia (44 participants), and seiz- ures (38 participants). Skeletal muscle (73 participants) and Leber hereditary optic neuropathy the gastrointestinal tract (26 participants) were also com- Twenty-eight participants were diagnosed with LHON; 23 monly affected (table 2). Brain imaging showed bilateral basal were males. Involvement of other organ-systems was com- monly reported. The most frequent extraocular affected organ ganglia involvement in 55 participants and/or lesions of the ff brainstem (27 participants). Variants in both nuclear and was the CNS (18 participants) with a constellation of di erent mitochondrial genomes were identified: 52 mtDNA and 45 manifestations including seizures (3 participants), migraine nDNA pathogenic variants in 21 different nuclear genes headaches (3 participants), hearing loss (2 participants), and (figure 1, table 2). ataxia (1 participant) (table e-1, links.lww.com/NXG/A226). Of 28 participants with a genetic diagnosis, only 1 had Myoclonus epilepsy with ragged red fibers a pathogenic nDNA variant; that participant was mis- Thirteen participants had a clinical diagnosis of myoclonus diagnosed before molecular genetic testing. epilepsy with ragged red fibers (MERRF). Six had childhood onset. Developmental problems were reported in 5 partic- Biochemical diagnosis fi ipants (table 2) and neuropathy in 4 (table e-2, links.lww. Biochemical de ciencies were recorded in 114 participants com/NXG/A226). Six participants had multiple lipomatosis. (table 1). Data on OxPhos activity were available in 666 fi All participants had a pathogenic variant in the mitochondrial individuals, of which defects were identi ed in 60. Combined fi MTTK gene, with the prototypical m.8344A>G pathogenic respiratory enzyme de ciencies were the most common ab- fi variant (figure 1) in 9 of them. normalities (21 participants). Isolated de ciencies in complex I or complex IV (cytochrome c oxidase [COX]) were the Mitochondrial encephalomyopathy, lactic acidosis, most common mono-enzymopathies. Disease onset was <2 and stroke-like episodes years in the majority of participants, except for individuals Seventy-one participants (28 males, 42 females, and 1 un- with COX deficiency whose onset spanned the entire age known) had MELAS. The majority had disease onset in spectrum (table 3). Molecular analysis revealed that isolated

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 2 Demographic, clinical, and genetic characteristics of the 264 North American Mitochondrial Disease Consortium Clinical Registry participants with canonical syndromes

LS MELAS cPEO/cPEO+ LHON MERRF

N = 97 N = 71 N = 55 N = 28 N = 13

Frequency (%) Frequency (%) Frequency (%) Frequency (%) Frequency (%)

Sexa

Female 48 (49.5) 42 (60.0) 37 (67.3) 5 (17.9) 7 (53.8)

Male 49 (50.5) 28 (40.0) 18 (32.7) 23 (82.1) 6 (46.2)

Missing — 1 ———

Age at onseta

<2 y 55 (61.8) 2 (3.1) — 1 (3.7) 4 (44.4)

≥2to<5y 23 (25.8) 1 (1.5) 4 (9.1) 1 (3.7) —

≥5 to <12 y 7 (7.9) 18 (27.7) 8 (18.2) 4 (14.8) 2 (22.2)

≥12 to <18 y — 11 (16.9) 8 (18.2) 4 (14.8) —

≥18 y 4 (4.5) 33 (50.8) 24 (54.5) 17 (63.0) 3 (33.3)

Missing 861114

Organ system involvement

CNS 87 (89.7) 68 (95.8) 33 (60.0) 18 (64.3) 13 (100.0)

Skeletal muscle 73 (75.3) 60 (84.5) 53 (96.4) 6 (21.4) 13 (100.0)

Heart 10 (10.3) 21 (29.6) 7 (12.7) 3 (10.7) 1 (7.7)

Developmental 91 (93.8) 22 (31.0) 2 (3.6) 3 (10.7) 5 (38.5)

Constitutionalb 34 (35.1) 39 (54.9) 15 (27.3) — 1 (7.7)

Psychiatric 10 (10.3) 23 (32.4) 19 (34.5) 4 (14.3) 4 (30.8)

PNS 7 (7.2) 11 (15.5) 10 (18.2) 2 (7.1) 4 (30.8)

Kidney 5 (5.2) 11 (15.5) 2 (3.6) 4 (14.3) —

Liver 1 (1.0) ————

Gastrointestinal 26 (26.8) 22 (31.0) 7 (12.7) 1 (3.6) 2 (15.4)

Genetic status

nDNA pathogenic variants 45 (46.4) 1 (1.4) 18 (32.7) 1 (3.6) —

mtDNA pathogenic variants 52 (53.6) 70 (98.6) 37 (67.3) 27 (96.4) 13 (100.0)

Abbreviations: cPEO = chronic progressive external ophthalmoplegia; LHON = Leber hereditary optic neuropathy; LS = Leigh syndrome; MELAS = mito- chondrial encephalomyopathy, lactic acidosis, and stroke-like episodes; MERRF = myoclonus epilepsy with ragged red fibers; mtDNA = mitochondrial DNA; nDNA = nuclear DNA; PNS = peripheral nervous system. a Missing values not included in the percentage denominator. b Includes cachexia, chronic fatigue, short stature (the patient’s height is below the 3rd percentile), and thinness (body mass index <18.5 kg/m2). complex I participants frequently had pathogenic variants in onset was <2 years in 38. PDC deficiency–associated pre- mtDNA (15/18 participants), whereas participants with COX sentations were diverse and included encephalopathy (14/44, deficiency commonly had nDNA pathogenic variants (9/14 31.8%), LS (9/44, 20.5%), encephalomyopathy (4/44, 9.1%), participants) (table 3). In both groups, the most frequent multisystemic syndrome (2/44, 4.5%), and cardiomyopathy clinical diagnosis was LS. Participants with combined OxPhos (1/44, 2.3%) (table 3). Forty-two participants had pathogenic deficiencies presented with a diverse spectrum of clinical di- variants in nDNA, predominantly in PDHA1 (35 partic- fi agnoses, with LS being the most common (table 3). ipants). CoQ10 de ciency was diagnosed in only 2 partic- ipants (of 220 tested). TP activity was reduced in 8 PDC activity was assessed in 213 participants, among whom participants of over 214 tested; all had pathogenic variants in defects were identified in 44 (28 females and 16 males). Age at TYMP.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Figure 1 Genes mutated in the canonical mitochondrial syndromes studied

Gene pathogenic variants in North American Mitochondrial Disease Consortium Registry par- ticipants with canonical syndromes. cPEO = chronic progressive external ophthalmoplegia; LHON = Leber hereditary optic neuropathy; MELAS = mitochondrial encephalomyopathy, lactic acidosis, and stroke-like episodes; MERRF = myoclonus epilepsy with ragged red fibers; mtDNA = mitochondrial DNA.

Molecular genetics SLEs (21.0% vs 7.1%, p < 0.0001), migraine headaches mtDNA pathogenic variants were more frequent than nDNA (19.6% vs 9.9%, p = 0.02), and dementia (9.2% vs 3.2%, p = (414 vs 252 participants). mtDNA variants were most prev- 0.04), with lower frequencies of upper motor signs (spasticity alent among older patients (38% of those with onset ≥18 and [6.8% vs 15.1%, p = 0.004] and hyperactive reflexes [5.6% vs 22% of those with onset <2), whereas nDNA variants were 13.1%, p = 0.017]). most common in the youngest group (50% of those with onset <2 [table 4]). The most common mtDNA pathogenic Participants with m.3243A>G variants were m.3243A>G in MT-TL1, which encodes The m.3243A>G pathogenic variant was reported in 138 tRNALeu(UUR) (138 participants) and single deletions (67 individuals, who were predominantly female (89 vs 48 males). participants: table e-2, links.lww.com/NXG/A226). Overall, Onset of symptoms frequently occurred when participants the distribution of pathogenic variants differed by age (p < were age >18 years (table e-2, links.lww.com/NXG/A226). 0.0001, table 4). The difference between the youngest and Ten different clinical syndromes were observed. The most oldest groups was particularly striking. Although nDNA var- frequent was MELAS (56 individuals), followed by diabetes iants were more than twice as frequent as mtDNA among and deafness (14 participants), encephalomyopathy (9 par- those with onset <2 (50.4% vs 22.2%), the reverse is true in ticipants), and maternally inherited diabetes (5 participants). the oldest-onset group (19.3% mtDNA vs 38.2% nDNA: post Multisystemic syndrome was diagnosed in 33 participants. hoc p < 0.0001). Pathogenic variants in 65 different nuclear CNS was affected in 127 (table e-2) participants, with seizure genes were reported (figure 2). Pathogenic variants in 38 (55 participants) and migraine (48 participants) as the most genes were reported uniquely in single participants (figure 2). common manifestations. Hearing loss was frequently recor- ded (92 participants), as well as diabetes mellitus that was The early age at onset of participants with nDNA pathogenic present in 57 participants. variants was associated with a higher frequency of de- velopmental delay (65.5%) compared with individuals with Participants with m.8344A>G mtDNA pathogenic variants (34.8%, p < 0.0001 (table 4). In Twenty-three participants had the m.8344A>G variant (10 contrast, patients with mtDNA pathogenic variants had males and 13 females) (table e-2, links.lww.com/NXG/ higher frequencies of endocrinopathies (24.6% vs 9.1%, p < A226). Although the pathogenic variant has been classically 0.0001) overall and diabetes mellitus in particular (19.1% vs associated with MERRF, 7 different phenotypes have been 2.8%, p = 0.0001). The overall frequencies of CNS manifes- identified in the NAMDC Registry. Nine participants had the tations were similar in participants with mtDNA and nDNA diagnosis of MERRF, whereas 9 individuals received other pathogenic variants (82.9% vs 86.1%, p = 0.37). However, clinical diagnoses. All participants had CNS involvement, with participants with mtDNA variants had higher frequencies of ataxia (15 participants) and myoclonus (16 participants) as

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 3 Frequencies of demographic, clinical, and genetic features of participants with specific biochemical diagnoses

Comp I Comp. II Comp III Comp. IV Comp V cRCE CoQ10 TP PDC

18/666 1/666 1/666 14/666 5/666 21/666 2/220 8/214 44/230

Sex

Female 91 1 8 3 121728

Male 9 ——6291116

Age at onset

<2 y 61 — 437— 138

≥2to<5y —— — 6152— 2

≥5 to <12 y 3 ——2 ———13

≥12 to <18 y 1 ——— —1 ———

≥18 y 4 ——218— 5 —

Clinical syndromes

Multisystemic syndrome 4 — 03 121— 2

Other clinical diagnosis —— 11 11——14

LS 71 — 5251— 9

Encephalomyopathy —— — 1 — 1 ——4

MELAS 1 ——— —2 ———

Myopathy —— — — — 1 ———

cPEO+ —— — 2 — 2 ———

Encephalopathy —— — — — 2 ——14

LHON 3 ——— —————

CPEO —— — — — 1 ———

KSS —— — — — 1 ———

SANDO 1 ——1 —————

MIDD 1 ——— —————

Alpers syndrome —— — 1 — 1 ———

MNGIE —— — — — ——8 —

Maternal inherited —— — — — ———— deafness

Cardiomyopathy —— — — — ———1

Hepatocerebral syndrome 1 ——— —2 ———

RIM with COX deficiency —— — — 1 ————

Genetic status

mtDNA pathogenic 15 — 15 5131— 2 variants

nDNA pathogenic variants 31 — 9 — 81842

Abbreviations: CoQ10 = coenzyme Q10; cPEO = chronic progressive external ophthalmoplegia; cRCE = combined respiratory chain enzymes; KSS = Kearns- Sayre syndrome; LHON = Leber hereditary optic neuropathy; LS = Leigh syndrome; MELAS = mitochondrial encephalomyopathy, lactic acidosis, and stroke- like episodes; MIDD = maternal inherited diabetes and deafness; MNGIE = mitochondrial neurogastrointestinal encephalopathy; mtDNA = mitochondrial DNA; nDNA = nuclear DNA; PDC = pyruvate dehydrogenase complex; RIM with COX deficiency = reversible infantile myopathy with cytochrome c oxidase deficiency; SANDO = sensory ataxic neuropathy with dysarthria and ophthalmoparesis; TP = thymidine phosphorylase.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 Table 4 Clinical features of participants with mtDNA pathogenic variants vs nDNA pathogenic variants

mtDNA pathogenic variants (N = 414) nDNA pathogenic variants (N = 252) p Value

Age at onset <0.0001

<2 y 78 (22.2) 120 (50.4)

≥2 and <5 y 40 (11.4) 43 (18.1)

≥5 and <12 y 66 (18.8) 10 (4.2)

≥12 and <18 y 33 (9.4) 19 (8.0)

≥18 y 134 (38.2) 46 (19.3)

Missing 63 14

Sexa 0.37

Male 183 (44.3) 102 (40.5)

Female 230 (55.7) 150 (59.5)

Missing 1 —

Organ system manifestations

CNSb 343 (82.9) 217 (86.1)

Seizures 126 (30.4) 101 (40.1) 0.16

Ataxia 126 (30.4) 98 (38.9) 0.15

SLE 87 (21.0) 18 (7.1) <0.0001

Migraine headaches 81 (19.6) 25 (9.9) 0.02

Myoclonic seizures 23 (5.6) 27 (10.7) 1.00

Dysarthria 68 (16.4) 56 (22.2) 0.36

Hearing loss 160 (38.6) 46 (18.3) <0.0001

Hyperactive reflexes 23 (5.6) 33 (13.1) 0.017

Other neurologic manifestation 135 (32.6) 99 (39.3) 0.73

Myoclonus 41 (9.9) 28 (11.1) 1.00

Dystonia 41 (9.9) 41 (16.3) 0.16

Dementia 38 (9.2) 8 (3.2) 0.042

Encephalopathy 37 (8.9) 39 (15.5) 0.15

Chorea 15 (3.6) 15 (6.0) 0.73

Spasticity 28 (6.8) 38 (15.1) 0.0042

Parkinsonism 5 (1.2) 7 (2.8) 0.73

PNSb 51 (12.3) 59 (23.4)

Axonal 39 (9.4) 42 (16.7) 0.0078

Demyelinating 18 (4.3) 24 (9.5) 0.0078

Ophthalmoparesis 75 (18.1) 62 (24.6) 0.09

Ptosis 126 (30.4) 70 (27.8) 0.48

DM 79 (19.1) 7 (2.8) <0.0001

Short stature 104 (25.1) 63 (25.0) 1.00

Myopathy 134 (32.4) 72 (28.6) 0.29

Developmental 144 (34.8) 165 (65.5) <0.0001

Continued

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 4 Clinical features of participants with mtDNA pathogenic variants vs nDNA pathogenic variants (continued)

mtDNA pathogenic variants (N = 414) nDNA pathogenic variants (N = 252) p Value

Anxiety 67 (16.2) 35 (13.9) 0.73

Depression 74 (17.9) 32 (12.7) 0.18

Bipolar disorder 9 (2.2) 6 (2.4) 0.79

Gastrointestinal 97 (23.4) 68 (27.0) 0.31

Endocrinec 102 (24.6) 23 (9.1) <0.0001

Abbreviations: DM = diabetes mellitus; mtDNA = mitochondrial DNA; nDNA = nuclear DNA; PNS = peripheral nervous system; SLE = stroke-like episode. a Missing values not included in the percentage denominator. b p Values are adjusted for 16 comparisons for CNS and 2 for PNS, using the Holm method.29 c Includes diabetes mellitus, hypothyroidism, hypogonadotropic, hypogonadism, hypoparathyroidism, and adrenal insufficiency. the most common manifestations (table e-2). Skeletal muscle and ophthalmoparesis (38 participants) were frequently reported weakness was reported in 21 participants. (table e-2). Hearing loss was less common (18 individuals).

Single large-scale mtDNA deletion (mtDNAD) Participants with POLG1 pathogenic variants mtDNAD was reported in 67 participants (29 males and 38 Seventy participants (25 males and 45 females) had pathogenic females) (table e-2, links.lww.com/NXG/A226) with a large variants in POLG1 (figure 2). Onset of symptoms occurred spectrum of clinical syndromes (10 different diagnoses), including from infancy through late adulthood (range <1–53 years). Kearns-Sayre syndrome (16 individuals), cPEO+ (15 individu- Twenty-four pathogenic variants were observed, distributed als), cPEO (12 individuals), and Pearson syndrome (9 individu- along the entire gene and with a high frequency in the linker als). Skeletal muscle (58/67, 86.6%) and CNS symptoms (44 region. Clinical presentation encompassed 11 different syn- participants) were prevalent (table e-2). Ptosis (47 participants) dromes. Sensory ataxic neuropathy with dysarthria and

Figure 2 Nuclear gene pathogenic variant frequency in the North American Mitochondrial Disease Consortium Registry

Each square represents 1 participant.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 ophthalmoparesis (SANDO) (18 participants) and Alpers participants are classified outside the classical mitochondrial syndrome (14 participants) were the most frequent. Ten had phenotypes with multisystemic syndrome (113 participants) SLEs. Seventeen participants had generic diagnoses of multi- the most common diagnosis. Sex stratification shows a slight systemic or other clinical syndrome. Of interest, phenotype preponderance of females, although the data do not include variability was observed in the 10 participants with the homo- asymptomatic carriers. Also noteworthy is the skewed racial zygous p.A467T pathogenic variant (2 Alpers-Huttenlocher composition. The majority of participants are Caucasian syndrome, 4 SANDO, 1 cPEO+, and 3 nonspecificclinical (>85%), whereas <3% were African Americans, and <4% were syndromes). Almost all the POLG1 participants (66/70) had Asian. In contrast, the 2010 US Census identified 72.4% CNS involvement. Skeletal muscle (61/70 participants) and Caucasian, 12.6% African American, and 4.8% Asian. Several peripheral nerve (33 participants) were often affected. The most factors may contribute to this disparity, and although we frequent CNS symptoms were ataxia (45/70 participants) and cannot exclude a recruiting bias, biological factors may play seizures (41/70 participants). a role in disease susceptibility of some populations.17

Participants with PDHA1 pathogenic variants Because MtDs are typically due to defects in energy metab- PDHA1 pathogenic variants were identified in 40 participants olism, the Registry includes data on biochemical defects. (11 males and 29 females) (figure 2). Age at onset was <2 years Enzyme activity assay helps to direct molecular analyses and in the majority (33/40 individuals). Developmental delay was to identify participants eligible for future therapies targeting observed in most participants (36/40), but information about specific enzymatic defects.18,19 Among the many such defects, this manifestation was unavailable for 3 participants. De- OxPhos deficiency has historically been considered the hall- velopmental delay was absent in only 1. Encephalopathy (13/40 mark of MtDs.7 In the Registry, the majority of the partic- participants) and LS (7/40 participants) were the most com- ipants had an OxPhos defect, followed by PDC deficiency. monly diagnoses. CNS was affected in 33 individuals, with seizure Participants with OxPhos defects most frequently presented (21/41 participants) being the most common manifestation. with multisystemic diseases. This may reflect the complex genetic control of the respiratory chain machinery.20,21 Re- Participants with OPA1 pathogenic variants markably few participants had complex V defects, possibly ffi Of the 13 participants with OPA1 pathogenic variants, 8 had due to low frequency, detection di culty, or both. optic atrophy, and 2 were diagnosed as optic atrophy type 1 (autosomal dominant optic atrophy Kjer type). One participant To portray the features of classic mitochondrial syndromes is was inappropriately given a clinical diagnosis of LHON before extremely important as new therapeutic strategies approach the mtDNA mutation was identified. Ten participants had the patient bedside. manifestations outside of the optic nerve including ptosis (4), ophthalmoparesis (4), hearing loss (4), seizures (3), ataxia (3), Participants with MELAS were prevalent in our cohort. Al- peripheral neuropathy (2), and dilated cardiomyopathy (1); though previous studies have suggested that male sex may hence, those patients were given non-autosomal dominant optic represent a risk factor for development of a full-blown ff atrophy diagnoses (multisystemic disease, cPEO plus, and MELAS syndrome, we did not observe a di erence by sex in encephalomyopathy). our population. Moreover, the ratio of males was similar among participants with the m.3243A>G variant (35%) and those with a diagnosis of MELAS (40%) (table e-2, links.lww. Discussion com/NXG/A226). In contrast to data obtained in other studies,22 MELAS was the most common diagnosis in par- Clinical registries are essential for collecting data on rare diseases; ticipants carrying the m.3243A>G variant. The high pro- those data can be critical for guiding their diagnoses and care. This portion of participants with MELAS in the NAMDC may be study, using data from the NAMDC Registry, is the first to depict due to ascertainment bias because some of the NAMDC sites the spectrum of mitochondrial disorders in North America. have research projects on this disorder.23

Although participants in the Registry can be enrolled based on Our data also confirm the relatively low prevalence of com- clinical manifestations (a genetic diagnosis is not required), this plete MERRF syndrome. Previous reports from the German analysis considers only participants with a definite genetic di- and Italian registries showed that only half of the participants agnosis. Nonetheless, it is worth highlighting the complexity of with the m.8344A>G pathogenic variant have classic MERRF the diagnostic pathway in these patients. Indeed, even in the era syndrome.24 The same proportion is observed in our cohort. of next-generation sequencing,13 and similarly to other large Of interest, the age at onset in our MERRF cohort is younger cohort studies,14,15 the NAMDC Registry shows a molecular than reported elsewhere, with an age at onset before 12 years diagnostic rate of 53.8% (722 participants of 1,341 received of age in 6 of 9 participants with available data. We identified a genetic diagnosis) (table e-3, links.lww.com/NXG/A226). hearing loss in 56.5% of our participants with the m.8344A>G variant and 46.2% with MERRF (table e-2, links.lww.com/ MtDs have diverse clinical presentations,16 which our analysis NXG/A226). This is in line with some other studies but lower confirms with 24 different clinical entities. Of interest, many than the German registry data.25,26

10 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG cPEO/cPEO+ diagnoses are common but less frequent than an initiative of the Office of Rare Diseases Research (ORDR), in previous reports.15 This discrepancy may have arisen National Center for Advancing Translational Sciences because the NAMDC Registry includes the multisystemic (NCATS). The project was funded by NIH U54 NS078059, syndrome category into which multiple complex cPEO+ sponsored by the NINDS, the Eunice Kennedy Shriver Na- participants were classified. tional Institute of Child Health and Development (NICHD), the Office of Dietary Supplements (ODS), and NCATS. In The Registry data highlight the vast genetic heterogeneity of addition, the NAMDC has received private funding from the the patients with MtDs. We observed a higher frequency of United Mitochondrial Disease Foundation (UMDF), a Pa- mtDNApathogenicvariantsovernDNA(table4).Thecom- tient Advocacy Group (PAG) for individuals with mito- bination of pediatric and adult clinics in the NAMDC facilitates chondrial disease and their family members, as well as the J. a comprehensive analysis across the MtD population. This may Willard and Alice S. Marriott Family Foundation. help explain why the proportions of mtDNA and nDNA var- iants in the Registry differ from those in other recent reports.27 Disclosure Other factors such as accessibility and feasibility of mtDNA Disclosures available: Neurology.org/NG. testing may also have an impact. In the Registry, participants with nDNA pathogenic variants had earlier disease onset than Publication history participants with mtDNA pathogenic variants, with onset be- Received by Neurology: Genetics June 13, 2019. Accepted in final form fore age 2 years in 50.4% individuals (table 4). We observed no December 16, 2019. differences in ptosis and ophthalmoparesis between the 2 groups, in contrast to what was observed in the Italian regis- try.15 High frequencies of SLE, diabetes mellitus, and deafness Appendix Authors in mtDNA participants were present, probably due to common mtDNA pathogenic variant-associated phenotypes. The higher Author Institution Contribution frequency of developmental abnormalities in the nDNA sub- Emanuele Department of Neurology, Data analysis, manuscript group confirms previous data27 and is concordant with the Barca, MD, Columbia University preparation, PhD Medical Center, NY communication among younger age at onset. canters, and study design

fi Yuelin Long, Department of Statistical analysis The identi cation in our cohort of several genes mutated only MS Biostatistics, Mailman in single individuals corroborates the ultra-rare nature of School of Public Health, Columbia University, New many MtDs genotypes. York

This study provides the first wide-ranging look at the MtD Victoria Department of Statistical analysis Cooley, MS Biostatistics, Mailman population in North America. The data also identify areas of School of Public Health, the Registry that require further improvement. Given the Columbia University, New York estimated prevalence of MtDs,28 it is likely that only a small proportion of North American participants has been cap- Robert Radboudumc, Nijmegen, Manuscript preparation Schoenaker, The Netherlands and data analysis tured. To address this limitation, the NAMDC has developed MD, BS a remote enrollment system to reach patients living in areas Valentina Department of Neurology, Data analysis, distant from a NAMDC center. In addition, the NAMDC has Emmanuele, Columbia University communication among created disease-specific natural history subregistries and other MD, PhD Medical Center, NY centers, and critical data studies that will improve the acquisition of longitudinal data review and help to identify outcome measures for future clinical Salvatore Department of Neurology, Data analysis, study DiMauro, MD Columbia University design, and critical data trials. Medical Center, NY review

Bruce H. Department of Pediatrics, Provided participant data Acknowledgment Cohen, MD Northeast Ohio Medical University and Akron The authors acknowledge the contributions of the Children’s Hospital participants and families who have participated in the Registry and of Amy Holbert, Health Informatics Institute/ Amel Karaa, Genetics Unit, Provided participant data MD Massachusetts General RDCRN, University of South Florida, for administrative Hospital, Boston support. In addition, they acknowledge the efforts of the Georgirene D. Department of Pediatrics, Provided participant data study coordinators. Vladutiu, PhD State University of New York at Buffalo

Richard Haas, Departments of Provided participant data Study funding MBBChir Neurosciences and The project was funded by NIH U54 NS078059. The North Pediatrics, University of American Mitochondrial Disease Consortium (NAMDC) is California at San Diego part of Rare Diseases Clinical Research Network (RDCRN), Continued

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 11 Appendix (continued) Appendix (continued)

Author Institution Contribution Author Institution Contribution

Johan L.K. Van Department of Pediatrics, Provided participant data Marni J. Falk, Mitochondrial Medicine Provided participant data Hove, MD, University of Colorado MD Frontier Program, Division PhD School of Medicine, of Human Genetics, The Aurora Children’sHospitalof Philadelphia and University Fernando Department of Molecular Provided participant data of Pennsylvania Perelman Scaglia, MD and Human Genetics, School of Medicine Baylor College of Medicine, Houston, TX; Texas Amy C. Mitochondrial Medicine Provided participant data Children’sHospital, Goldstein, MD Frontier Program, Division Houston; Joint BCM-CUHK of Human Genetics, The Center of Medical Genetics, Children’s Hospital of Prince of Wales Hospital, Philadelphia and ShaTin, New Territories, University of Pennsylvania Hong Kong Perelman School of Medicine Sumit Parikh, Department of Neurology, Provided participant data MD Cleveland Clinic, OH Mark Department of Neurology, Provided participant data Tarnopolsky, McMasters University, Jirair K. Departments of Genetics Provided participant data MD, PhD Toronto, ON, Canada Bedoyan, MD, and Genome Sciences and PhD Pediatrics, and Center for Andrea Children’s National Provided participant data Human Genetics, University Gropman, MD Medical Center Hospitals Cleveland Medical Center, Case Kathryn Office of Dietary Critical data review Western Reserve Camp, MS, RD Supplements, National University, OH Institutes of Health, Bethesda, MD Susanne D. Departments of Genetics Provided participant data DeBrosse, MD and Genome Sciences and Danuta Eunice Kennedy Shriver Critical data review Pediatrics, and Center for Krotoski, PhD National Institute of Child Human Genetics, University Health and Human Hospitals Cleveland Medical Development, National Center, Case Western Institutes of Health, Reserve University, OH Bethesda, MD

Ralitza H. Departments of Provided participant data Kristin Department of Neurology, Provided participant data Gavrilova, MD Neurology and Clinical Engelstad, MS Columbia University Genomics, Mayo Clinic, Medical Center, NY Rochester, MN Xiomara Q. Department of Neurology, Provided participant data Russell P. Department of Neurology, Provided participant data Rosales, MD Columbia University Saneto, DO, University of Washington, Medical Center, NY PhD Seattle Children’s Hospital Joshua Kriger, Department of Statistical analysis Gregory M. Department of Pediatrics, Provided participant data MS Biostatistics, Mailman Enns, MBChB Stanford University, Palo School of Public Health, Alto, CA Columbia University, New York Peter W. Department of Medicine, Provided participant data Stacpoole, University of Florida at Johnston Department of Provided participant data MD, PhD Gainesville Grier, MS Biostatistics, Mailman School of Public Health, Jaya Ganesh, Department of Pediatrics, Provided participant data Columbia University, New MD Cooper University York Hospital, Camden, NJ Richard Department of Database curation and Austin Department of Pediatrics, Provided participant data Buchsbaum Biostatistics, Mailman data set preparation Larson, MD University of Colorado School of Public Health, School of Medicine, Columbia University, New Aurora York

Zarazuela Mitochondrial Medicine Provided participant data John L.P. Department of Statistical analysis, Zolkipli- Frontier Program, Division Thompson, Biostatistics, Mailman manuscript preparation, Cunningham, of Human Genetics, The PhD School of Public Health, and study design MD Children’s Hospital of Columbia University, New Philadelphia and York University of Pennsylvania Perelman School of Michio Department of Neurology, Data analysis, manuscript Medicine; University of Hirano, MD Columbia University preparation, Pennsylvania Perelman Medical Center, NY communication among School of Medicine, centers, study design, and Philadelphia critical data review

12 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG References 16. Zolkipli-Cunningham Z, Xiao R, Stoddart A, et al. Mitochondrial disease patient 1. DiMauro S, Davidzon G. Mitochondrial DNA and disease. Ann Med 2005;37: motivations and barriers to participate in clinical trials. PLoS One 2018;13:e0197513. 222–232. 17. Salvatore DiMauro CP. Mitochondrial disorders due to mutations in the mitochon- ’ 2. McFarland R, Taylor RW, Turnbull DM. A neurological perspective on mitochondrial drial genome. In: Rosenberg RN, Pascual JM, editors. Rosenberg s Molecular and disease. Lancet Neurol 2010;9:829–840. Genetic Basis of Neurological and Psychiatric Disease. 5th Edition. San Diego, CA: – 3. Chinnery PF. Mitochondrial disorders overview. In: Adam MP, Ardinger HH, Pagon Elsevier Science Publishing Co Inc. 2015:271 281. RA, et al, editors. GeneReviews®. Seattle, WA: University of Washington, Seattle; 2014. 18. Quinzii CM, Emmanuele V, Hirano M. Clinical presentations of coenzyme q10 fi – 4. Gorman GS, Schaefer AM, Ng Y, et al. Prevalence of nuclear and mitochondrial DNA de ciency syndrome. Mol Syndromol 2014;5:141 146. mutations related to adult mitochondrial disease. Ann Neurol 2015;77:753–759. 19. Halter JP, Michael W, Schupbach M, et al. Allogeneic haematopoietic stem cell 5. Thompson R, Robertson A, Lochm¨uller H. Natural history, trial readiness and gene transplantation for mitochondrial neurogastrointestinal encephalomyopathy. Brain – discovery: advances in patient registries for neuromuscular disease. Adv Exp Med Biol 2015;138:2847 2858. ffi 2017;1031:97–124. 20. Taylor RW, Pyle A, Gri n H, et al. Use of whole-exome sequencing to determine the 6. Louis ED, Mayer SA, Rowland LP. Merritt’s Neurology. Philadelphia, PA: Lippincott genetic basis of multiple mitochondrial respiratory chain complex deficiencies. JAMA Williams & Wilkins; 2016. 2014;312:68–77. 7. DiMauro S, Schon EA, Carelli V, Hirano M. The clinical maze of mitochondrial 21. Kemp JP, Smith PM, Pyle A, et al. Nuclear factors involved in mitochondrial trans- neurology. Nat Rev Neurol 2013;9:429–444. lation cause a subgroup of combined respiratory chain deficiency. Brain 2011;134: 8. Wallace DC, Zheng XX, Lott MT, et al. Familial mitochondrial encephalomyopathy 183–195. (MERRF): genetic, pathophysiological, and biochemical characterization of a mito- 22. Nesbitt V, Pitceathly RD, Turnbull DM, et al. The UK MRC Mitochondrial Disease Patient chondrial DNA disease. Cell 1988;55:601–610. Cohort Study: clinical phenotypes associated with the m.3243A>G mutation–implications 9. Hirano M, Ricci E, Koenigsberger MR, et al. MELAS: an original case and clinical for diagnosis and management. J Neurol Neurosurg Psychiatry 2013;84:936–938. criteria for diagnosis. Neuromuscul Disord 1992;2:125–135. 23. Kaufmann P, Engelstad K, Wei Y, et al. Natural history of MELAS associated with 10. Wallace DC, Singh G, Lott MT, et al. Mitochondrial DNA mutation associated with mitochondrial DNA m.3243A>G genotype. Neurology 2011;77:1965–1971. Leber’s hereditary optic neuropathy. Science 1988;242:1427–1430. 24. Altmann J, Buchner B, Nadaj-Pakleza A, et al. Expanded phenotypic spectrum of the 11. Cohen BH, Chinnery PF, Copeland WC. POLG-related disorders. In: Adam MP, m.8344A>G “MERRF” mutation: data from the German mitoNET registry. J Neurol Ardinger HH, Pagon RA, et al, editors. GeneReviews®. Seattle, WA: University of 2016;263:961–972. Washington, Seattle; 1993. 25. Chinnery PF, Howell N, Lightowlers RN, Turnbull DM. MELAS and MERRF. The 12. Hirano M. Mitochondrial neurogastrointestinal encephalopathy disease. In: Adam relationship between maternal mutation load and the frequency of clinically affected MP, Ardinger HH, Pagon RA, et al, editors. GeneReviews®. Seattle, WA: University of offspring. Brain 1998;121:1889–1894. Washington, Seattle; 1993. 26. Mancuso M, Orsucci D, Angelini C, et al. Phenotypic heterogeneity of the 8344A>G 13. McCormick E, Place E, Falk MJ. Molecular genetic testing for mitochondrial disease: mtDNA “MERRF” mutation. Neurology 2013;80:2049–2054. from one generation to the next. Neurotherapeutics 2013;10:251–261. 27. Witters P, Saada A, Honzik T, et al. Revisiting mitochondrial diagnostic criteria in the 14. Pronicka E, Piekutowska-Abramczuk D, Ciara E, et al. New perspective in diagnostics new era of genomics. Genet Med 2018;20:444–451. of mitochondrial disorders: two years’ experience with whole-exome sequencing at 28. Skladal D, Halliday J, Thorburn DR. Minimum birth prevalence of mitochondrial a national paediatric centre. J Transl Med 2016;14:174. respiratory chain disorders in children. Brain 2003;126:1905–1912. 15. Orsucci D, Angelini C, Bertini E, et al. Revisiting mitochondrial ocular myopathies: 29. Holm S. A simple sequentially rejective multiple test procedure. Scand J Statist 1979; a study from the Italian network. J Neurol 2017;264:1777–1784. 6:65–70.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 13 ARTICLE OPEN ACCESS Characterization of the phenotype with cognitive impairment and protein mislocalization in SCA34

Marie Beaudin, MD, MSc, Leila Sellami, MD, MSc, Christian Martel, MSc, Lydia Touzel-Deschˆenes, MSc, Correspondence ´ Gabrielle Houle, BSc, Laurence Martineau, MD, Kevin Lacroix, MD, Andr´eane Lavall´ee, MD, Dr. Dupre [email protected] Nicolas Chrestian, MD, Guy A. Rouleau, MD, PhD, François Gros-Louis, PhD, Robert Jr Laforce, MD, PhD, and Nicolas Dupr´e, MD, MSc

Neurol Genet 2020;6:e403. doi:10.1212/NXG.0000000000000403 Abstract Objective To better characterize the neurologic and cognitive profile of patients with spinocerebellar ataxia 34 (SCA34) caused by ELOVL4 mutations and to demonstrate the presence of ELOVL4 cellular localization and distribution abnormalities in skin-derived fibroblasts.

Methods We investigated a 5-generation French-Canadian kindred presenting with a late-onset cerebellar ataxia and recruited age- and education-matched controls to evaluate the presence of neuro- cognitive impairment. Immunohistochemistry of dermal fibroblasts derived from a patient’sskin biopsy was performed.

Results Patients had a late-onset slowly progressive cerebellar syndrome (mean age at onset 47 years; range 32–60 years) characterized by truncal and limb ataxia, dysarthria, hypometric saccades, and saccadic pursuits. No patient had past or current signs of erythrokeratodermia variabilis, which had previously been reported. MRI revealed cerebellar atrophy, with pontine atrophy (4 of 6 patients), and cruciform hypersignal in the pons (2 of 6 patients). Fluorodeoxyglucose- PET showed diffuse cerebellar hypometabolism in all 5 tested patients with subtle parietal hypometabolism in 3. Significant cognitive deficits were found in executive functioning, along with apparent visuospatial, attention, and psychiatric involvement. Immunohistochemistry of dermal fibroblasts showed mislocalization of the ELOVL4 protein, which appeared punctate and aggregated, supporting a dominant negative effect of the mutation on protein localization.

Conclusions Our findings support the pathogenicity of ELOVL4 mutations in cerebellar dysfunction and provide a detailed characterization of the SCA34 phenotype, with neurocognitive changes typical of the cerebellar cognitive-affective syndrome.

From the Department of Medicine (M.B., L.S., N.D.), Faculty of Medicine, Universit´e Laval; Division of Neurosciences (M.B., L.M., N.D.), CHU de Qu´ebec – Universit´e Laval; Clinique Interdisciplinaire de M´emoire (L.S., R.L.), CHU de Qu´ebec; Laval University Experimental Organogenesis Research Center/LOEX (C.M., L.T.-D., F.G.-L.), Division of Regenerative Medicine, CHU de Qu´ebec Research Center – Enfant-J´esus Hospital; Montreal Neurological Institute (G.H., G.A.R.), McGill University, Qu´ebec, Canada; CHU Grenoble-Alpes (L.M.), Grenoble, France; CIUSSS de la Mauricie-et-du-Centre-du-Qu´ebec (K.L.), Trois-Rivi`eres; Centre universitaire d’ophtalmologie (A.L.), Department of Surgery, Faculty of Medicine, CHU de Qu´ebec – Universit´e Laval; and Centre M`ere-Enfant-Soleil (N.C.), Universit´e Laval, Qu´ebec, Canada.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by CHU de Quebec - Universit´e Laval. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary CCAS = cerebellar cognitive-affective syndrome; DTR = deep tendon reflex; EKV = erythrokeratodermia variabilis; FBS = fetal bovin serum; FDG-PET = fluorodeoxyglucose-PET; MoCA = Montreal Cognitive Assessment; RCFT = Rey Complex Figure Test; SCA34 = spinocerebellar ataxia 34; STGD3 = Stargardt-like macular dystrophy type 3; VLCFA = very long chain fatty acid; VLC-SFA = very-long-chain saturated fatty acid.

Spinocerebellar ataxia 34 (SCA34),alsocalledataxia(ATX)- the assessment and rating of ataxia (SARA). MRI was per- ELOVL4 according to the revised nomenclature,1 was first formed in 6 affected individuals, and 3 underwent nerve con- described in a French-Canadian family with ataxia and eryth- duction studies. All patients were specifically questioned rokeratodermia variabilis (EKV)2,3 caused by a dominant mu- regarding the presence of cutaneous involvement, and one self- tation in ELOVL4, which encodes the elongation of very-long- reporting patient had a thorough evaluation by a dermatologist. chain fatty acid (VLCFA) protein 4 (ELOVL4). Since then, 2 Japanese families presenting with a cerebellar disorder and Five patients and as many cognitively healthy controls un- multisystem atrophy–like features on MRI were reported,4 derwent formal neurocognitive and neuropsychiatric assess- along with 2 sporadic patients with neurocutaneous ment at the CHU de Qu´ebec Memory Clinic (cliniq involvement.5,6 The disease phenotype is still poorly charac- uedememoire.ca). The other affected participants did not terized, and little is known on its impact on other neurologic wish to collaborate beyond the standard clinical evaluation. functions, including cognition. Controls were family caregivers and were age- and education- matched to participants. The Mini-Mental State Examination Cognitive and affective dysfunction caused by cerebellar dis- was performed with each participant. Data derived from the ease has been labeled the cerebellar cognitive-affective syn- Montreal Cognitive Assessment12 and additional standard- drome (CCAS), which is characterized by impairment in ized and local cognitive measures were summarized according executive function, visuospatial organization, language, and to composite scores for the main cognitive domains: attention psychiatric features.7,8 Neurocognitive impairment has been (digit span and numeric substitution), executive functions – reported in selected hereditary ataxias,9 11 but previous reports (abstract thinking, lexical verbal fluency, Go/No-Go, Luria, – of patients with SCA34 only reported normal cognition.3 6 Trail Making B, and Rey Complex Figure Test [RCFT] copy), visuospatial skills (block copy and clock drawing), To date, the biochemical validation of ELOVL4 dysfunction memory (5-word list free and cued recall), and language in SCA34 remains elusive. Previous reports have demon- (naming, comprehension, and sentence repetition tasks) strated that serum levels and ratios of VLCFAs (C22 to C26), (appendix e-1, links.lww.com/NXG/A228, for detailed tasks corresponding to the ELOVL4 substrates and products, included in each domain). Affective symptoms, including remained within normal limits.3,4 mood alterations and anxiety, along with behavioral changes were assessed by asking patients and caregivers about relevant Our primary objective was to characterize the neurologic and neuropsychiatric symptoms and by observation of affect and neurocognitive phenotype of patients with SCA34 from appropriateness of behavior during the cognitive evaluation. a large family to better define disease manifestations. The Patients’ scores were compared with controls’ using the paired secondary objective was to demonstrate that skin fibroblasts sample t test. The cognitive involvement appeared milder of patients with the mutated ELOVL4 allele had subcellular in the youngest patient (V-1), so she was referred for a de- localization or distribution anomalies supporting ELOVL4 tailed neuropsychological assessment (California Verbal dysfunction as the cause of SCA34. Learning Test-II, Delis-Kaplan Executive Function System, DO-80, Taylor complex figure, Minnesota Multiphasic Per- sonality Inventory, Ruff 2 and 7, Tower of London test, Methods and Wechsler Adult Intelligence Scale 4th edition). Func- tional imaging with [18F]-fluorodeoxyglucose PET (FDG- Clinical and cognitive assessment PET) was performed in the 5 patients who underwent cog- We investigated a large 5-generation French-Canadian kindred nitive evaluation. presenting with a late-onset pure cerebellar ataxia with ponto- cerebellar atrophy on MRI. A total of 19 affected and unaffected individuals provided written informed consent to participate in Whole-exome sequencing and genetic analysis the study. We collected genomic DNA and conducted clinical To identify the genetic mutation underlying ataxia in this interviews and neurologic evaluations for all participants. All family, whole-exome sequencing was performed on 5 affected participants were evaluated in December 2013, and 6 affected members. Genomic DNA was extracted from peripheral blood patients had regular clinical follow-up until July 2018 when they leukocytes following the manufacturer’s protocol (Puregene; were evaluated by an ophthalmologist to search for retinal ab- Gentra Systems, Inc). Whole-exome sequencing (WES) was normalities and were formally rated according to the scale for performed on 5 affected members of the family. Exome capture

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG was performed using SeqCap EZ Exome Library v3 kit (Nim- Standard protocol approvals, registrations, blegen, Roche) with 1 μg of genomic DNA. Exon-enriched and patient consents DNA libraries from these 5 individuals were sequenced on the The research ethics committee of the CHU de Qu´ebec ap- Illumina HiSeq 2000 with a paired-end 100-bp length config- proved the participation of human subjects to this study (PEJ- uration. Reads were mapped against the Human Reference 299, 2012-1311), and written informed consent was obtained Genome hg19 using the Burrows-Wheeler Aligner (v0.7.5). from all participants. The participants received no financial Single nucleotide variations and copy number variations, in- compensation. cluding insertions/deletions (indels), were called using the 13 Genome Analysis Toolkit (v2.6) and annotated using Data availability 14 ANNOVAR13. The exomes were analyzed using step-by-step Anonymized data not published in the article will be shared by filtering strategies. Basic filters were used to remove low-quality request from any qualified investigator. variants: minimum position coverage >6, minimum reads sup- porting the variant >3, and minimum mutation frequency >0.15. Genetic variants were then filtered to retain only variants Results respecting the following criteria: (1) shared by all 5 affected family members sequenced by WES, (2) nonsynonymous sin- Genetic results gle nucleotide variants, splicing site variants, or coding indels, A total of 19 individuals (9 affected and 10 unaffected) un- (3) with a minor allele frequency <0.1% in Exome Aggregation derwent a complete neurologic evaluation and genetic testing Consortium, and (4) not present in 186 controls from our in- (table 1). Routine biochemical investigations were normal, and house data set. To validate our results and examine segregation initial genetic screening had excluded SCA1, 2, 3, 6, 7, 8, 17, and of the identified ELOVL4 DNA variations with the disease in fragile X syndrome for this family. WES in 5 affected individuals thefamily,weusedspecific PCR primer pairs (forward: revealed 426,715 DNA variations. Only 1 variant, a heterozy- CATTGCTTTCCACTGAACACA; reverse CATGCCTTG- gous missense mutation in the ELOVL4 gene (NM_022726.3 TACATTTTGTGC) to amplify DNA from 18 family members c.504G>C, p.L168F), fulfilled the filtering criteria. The L168F and sequenced them by Sanger sequencing (Applied Bio- amino acid substitution was located at a highly conserved res- system’s 3730xl DNA Analyzer technology). One additional idue in the third transmembrane domain of the protein. This family member was tested with commercially available Sanger mutation was predicted to be possibly damaging with a score of sequencing of ELOVL4 from Prevention Genetics. 0.867 according to the PolyPhen-2 web site (genetics.bwh. harvard.edu/pph2/).16 This mutation was confirmed in 4 ad- Fibroblast cell isolation and culture ditional affected family members, but was absent in the 10 A skin biopsy was performed on the patient who reported cu- unaffected relatives. Hence, perfect segregation of the identified taneous involvement. The fibroblast cells were isolated from c.504G>C mutation with the disease status in the family was skin biopsies and kept in liquid nitrogen until used as previously observed (figure e-1, links.lww.com/NXG/A228). described.15 Briefly, fibroblasts were cultivated in Dulbecco's Modified Eagle Medium with 10% fetal bovin serum (Seradigm) This variant had already been reported to cause SCA34 in 3 with 100 IU/mL penicillin G (Sigma-Aldrich) and 25 μg/mL another French-Canadian family, but no common ancestor fi gentamicin (Gemini Bio-Products) in 8% CO2 at 37°C. They could be identi ed by family history and the haplotype was were cultured in flasks until confluency (100%) prior to be split not reconstructed from the exome data. With agreement of and seeded in a 6-well plate containing sterile glass coverslips. the authors, exomes from the previously described family were retrieved, and kinship coefficients were calculated using Immunofluorescence analyses KING.17 All pairs of sample between the 2 studies had a phi Fibroblasts cells, grown on glass coverslips, were fixed in 4% score of <0.03, which corresponds to a third-degree re- (wt/vol) paraformaldehyde in phosphate-buffered saline for lationship or second-degree cousins. Although calculating 20 minutes and blocked for 1 hour in 5% (vol/vol) goat relatedness coefficients using exome sequencing data can be serum 0.1% Triton X-100 in phosphate-buffered saline at biased due to sparse and unevenly distributed data, this room temperature. Immunofluorescence was performed suggested that the patients from the 2 families were not close using antibodies recognizing ELOVL4 (rabbit polyclonal, relatives. Considering the close geographic proximity and Abcam ab224608) and calnexin (mouse monoclonal, Invi- common French-Canadian ethnical background, it remains trogen MA3-027) and used at 1:200 diluted in blocking likely that both families have a distant common ancestor solution overnight at 4°C. Fluorescent signal distribution that was not identified by family history or relatedness was visualized by epifluorescence microscopy after in- coefficient. cubation with secondary antibodies, anti-mouse or anti- rabbit IgG conjugated to Alexa Fluor 488 (green) or 594 Clinical assessment (red) (Invitrogen mouse 488 #A11001 and rabbit 594 The disorder was characterized by a late-onset cerebellar syn- #A21207) diluted 1:500 for 2 hours at room tempera- drome (mean age at onset 47 years; range 32–60 years) with ture, and mounted in Prolong (Fluoromount-G with 4',6- slow progression. Patients generally required a walker in their Diamidino-2-phenylindole, Invitrogen #00-4959-52). 70s and became a wheelchair user in their 80s. Truncal and gait

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 ataxia, along with dysmetria and dysarthria, were remarkable in all patients (table 1). The median SARA score was 16 at last evaluation (range 8–21.5), and there was a correlation between SARA score and age with a Pearson correlation coefficient of dDTR and EPR 0.84, suggesting disease progression with advancing age. Hypometric saccades and saccadic pursuits were remarkable on ) * ( examination. Most patients denied current or previous der- matologic involvement, but 1 patient (IV-18) reported ery- thematous dry skin lesions on the lower limbs during winter beginning in her fifth decade (figure e-2, links.lww.com/NXG/ A228). Nummular dermatitis was diagnosed, but there was no typical EKV lesion. Two other patients reported occasional dry skin on the lower limbs in winter, but none had erythematous lesions, and there were no active lesions at the time of evalu- ation. Moreover, 3 patients reported a history of severe gingi- vitis that occurred in the second decade and caused

ramidal rigidity; N = normal; NA = not available; ND = nummular a generalized edentation that required dentures. One patient (V-1) had bilateral pale pisciform perimacular retinal lesions on N LOVPT SMPN Slow* NN NNN N N N MPN Slow** Slow* ophthalmologic Slow* Slow UI examination. She had no complains of visual

(sensitivo)motor polyneuropathy; UI = urinary incontinence. loss and was examined by a retinal disease specialist who concluded that there was no evidence of Stargardt disease or − − − − − retinitis pigmentosa. All other evaluated patients had a normal ophthalmologic examination. Three patients had mild sensory deficits on evaluation, 2 of whom also had decreased deep tendon reflexes (DTRs). Polyneuropathy was confirmed in 2 of 3 tested patients (table 1). + + N N N Slow Hypometric saccades and saccadic pursuit Tremor Strength Sensibilities EMG/NCS Evolution Others MRI was performed in 6 patients and revealed mild (1/6) to severe (5/6) vermian and hemispheric cerebellar atrophy in all patients along with pontine atrophy in 4 (figure 1A). Cruciform hypersignal in the pons, also called the hot cross

− bun sign, was present in 2 patients (III-27 and IV-3) (figure fi

mutation 1B). Other ndings were reported only in selected patients, notably mild diffuse cerebral cortical (2/6) or subcortical (2/6) atrophy and leukoencephalopathy (2/6). FDG-PET ELOVL4 revealed diffuse cerebellar hypometabolism in all 5 patients. Three patients also had a very subtle bilateral (2/5) or right (1/5) parietal hypometabolism (figure 2C). One other pa- tient had heterogeneous cerebral hypometabolism with pre- served basal ganglia, brainstem, and mesiotemporal regions, which was possibly secondary to diffuse vascular involvement. In this patient, MRI had shown subcortical atrophy, but no evidence of previous stroke. Limb and truncal ataxia Dysarthria SARA score Nystagmus Cognitive assessment Five patients were evaluated at the Clinique Interdisciplinaire de M´emoire du CHU de Qu´ebec. Their performance was compared with 5 age- and education-matched controls (mean education for patients = 12.4 ± 1.51 years vs 13.4 ± 2.07 for controls). The cognitive involvement was homogeneous within affected patients with significant deficits in executive func- tioning and apparent impairment in visuospatial skills and at- tention that did not reach statistical significance (table 2).

Detailed neurologic findings in affected individuals with Working memory, psychomotor speed, set shifting, inhibition, 50 86 +++ ++ 20.5 + + 50 83 +++ + NA + + NA4540 9350 7950 7432 72 66 +++ 44 + ++ +++ + ++ ++ NA ++ + + NA ++ NA 21.5 + 16 + 11 + NA 8 + + + + NA + + NA NA + N Slow** LOV SCI Slow UI, ND, and dDTR 60 82 +++ ++ 16.5 + + + N LOT Slow* Reported age at onset Age at last evaluation and planning skills were particularly affected as revealed by failure to perform backward digit span, decreased lexical verbal fl III-7 III-20 III-24 III-27 IV-3 IV-6 IV-18 V-1 III-3 Table 1 ID Abbreviations: + = present/mild; ++ = moderate; +++ = severe; * = uses a walker;** = uses a wheelchair; dDTR = decreased deep tendon reflex; EPR = extrapy dermatitis; LO(V)(P)(T) = loss of vibration, proprioception, temperature; SCI = severe cognitive impairment with decreased cooperation; (S)MPN = uency, and impaired trail making, Luria sequence, Go/No-Go,

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 1 Neuroimaging findings in patients with the c.504G>C mutation in ELOVL4

(A) Sagittal MRI of a 68-year-old man (IV-3) showing severe cerebellar at- rophy with pontine atrophy. (B) Axial MRI from the same patient showing T2 cruciform hypersignal in the pons (indicated by red arrow). (C) Fluo- rodeoxyglucose-PET of a 70-year-old woman (IV-6) showing severe cere- bellar hypometabolism and minimal biparietal hypometabolism. Views are indicated in the top row: Lat. = lateral; Sup. = Superior; Inf. = Inferior; Ant = Anterior; Post. = Posterior; Med. = Medial. GLB indicates normalized to the global metabolism of the brain. and RCFT (figure 2). Visuoconstructive skills and attention youngest patient (V-1) revealed a milder impairment in se- were also impaired in all patients, but this did not reach sta- lective attention and working memory. Psychiatric features tistical significance. Patients did not present deficits in memory were remarkable in 3 patients, 2 of whom exhibited frank dis- and language. Detailed neuropsychological testing in the inhibition and euphoria, whereas impulsiveness and anxiety

Figure 2 Impaired copy of the RCFT in 3 patients with the c.504G>C mutation in ELOVL4

(A) RCFT model. (B) Sixty-four-year-old woman with 12 years of education (IV-18). (C) Seventy- two-year-old man with 11 years of education (IV-3) shows significant juxtaposition of details. (D) Eighty-year-old man with 21 years of edu- cation (III-3) with a severe dysexecutive syn- drome. RCFT = Rey Complex Figure Test.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 com/NXG/A228). Notably, decreased DTR and peripheral Table 2 Results of the cognitive evaluation in patients neuropathy were both observed in 2 patients from the present with SCA34 and controls study, whereas 4 of 8 patients from the previously reported 3 Patients Controlsa French-Canadian family had a mild peripheral neuropathy. Clinical domains (n = 5) (n = 5) p Valueb However, another study reported increased DTR and positive fi Cognitive—mean Babinski signs, upper motor neuron ndings that may be as- ± SD sociated with the multiple system atrophy–like findings on 4 Attention 5.6 ± 1.5 6.8 ± 0.4 0.138 imagery in these patients. Of interest, 2 patients from the present study also presented with a hot cross bun sign, but Executive 2.4 ± 2.1 5.8 ± 0.4 0.042 DTR and plantar reflexes were normal. Hence, it appears that Visuospatial 9.2 ± 1.1 10.6 ± 0.5 0.053 the cerebellar syndrome may be associated with peripheral polyneuropathy or pyramidal signs in patients with SCA34. Memory 4.2 ± 0.4 4.2 ± 0.4 NA

Language 3±0 3±0 NA We report patients with SCA34 with deficits in executive ff Psychiatric—no. functioning that were statistically di erent from age- and (%) education-matched controls, along with apparent visuospatial fi Disinhibition 2 (40) 0 and attention de cits. Although some of the cognitive tasks could have been affected by reduced hand coordination, such as Euphoria 2 (40) 0 the RCFT and Trail Making Test, the deficits observed went Impulsiveness 3 (60) 0 beyond what would be expected in a pure cerebellar motor syndrome. This was associated with psychiatric features in 3 Anxiety 1 (20) 0 patients. These findings appear typical of the CCAS7,8 and Panic 1 (20) 0 provide additional evidence of neurocognitive and neuro- behavioral impairment in hereditary degenerative ataxias. This Abbreviation: SCA34 = spinocerebellar ataxia 34. a Controls are age- and education-matched to patients. is consistent with the results of recent functional imaging b p Values are calculated with the paired sample t test. studies showing that the cerebellum is composed of several functional areas: the sensorimotor region in the anterior lobe and lobule VIII, the cognitive region in the posterior lobes, and – were noted in a third patient (see appendix e-2, links.lww.com/ the affective region in the posterior vermis.18 20 The CCAS is NXG/A228, for detailed evaluations). expected to result from disruption of cerebro-cerebellar loops, particularly cerebellar projections to association cortices in the Aberrant ELOVL4 localization in dorsolateral prefrontal and parietal lobes and to the limbic patient fibroblasts – regions.19 21 In our patients, functional imaging showed cere- Immunofluorescence analysis of fibroblasts from a healthy bellar hypometabolism, which was associated with mild parietal control showed homogeneous staining of ELOVL4 seemingly hypometabolism in 3 patients. This could be attributable to colocalizing with the endoplasmic reticulum marker calnexin at a diaschisis phenomenon, wherein severe damage to the cere- the perinuclear space (figure 3A). However, mislocalization of bellar hemispheres could result in reduced cerebral metabolic the ELOVL4 protein beyond the perinuclear region and ab- activity through deactivation of the cerebello-thalamic-cortical normal punctates and aggregates can be observed in patient- projections.22 Indeed, the parietal lobe has important afferent derived fibroblasts carrying the c.504G>C mutation (figure 3B). and efferent connections with the cerebellum. These appear to be involved in the pathophysiology of parietal ataxia, or crossed Discussion cerebellar diaschisis, a phenomenon in which patients with a primary parietal lesion develop secondary cerebellar dys- We report a large family with a c.504G>C mutation in ELOVL4 function.23 The parietal hypometabolism observed here could associated with cerebellar ataxia and cognitive impairment with reflect a similar phenomenon, i.e., parietal dysfunction sec- prominent executive dysfunction. Our findings provide addi- ondary to a primary cerebellar lesion. A previous report of tional support to the putative role of ELOVL4 mutations in parietal hypometabolism in a pure recessive cerebellar ataxia cerebellar dysfunction. Moreover, we report evidence of phenotype caused by SYNE1 recessive mutations supports this ELOVL4 cellular abnormalities in SCA34 by demonstrating hypothesis.9 This phenomenon may contribute to the visuo- mislocalization and abnormal aggregation of the protein in spatial and attention deficits observed in our patients and may apatient’s dermal fibroblasts. be amenable to potential treatments targeting diaschisis.22

In our cohort, all patients presented with a prominent cere- As opposed to previously reported patients with the same bellar syndrome of late onset and slow progression. So far, mutation,3 our patients did not report any history of skin ELOVL4-linked SCA34 has been reported in patients of Ca- lesions associated with EKV. All patients were thoroughly nadian, Japanese, and Brazilian origins, and clinical features vary questioned regarding past or current erythematous or hyper- significantly among described patients (table e-1, links.lww. keratotic lesions, but only the patient reporting active lesions

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 3 Immunofluorescence staining of dermal fibroblasts from a healthy control and a patient with SCA34

(A) ELOVL4 staining in healthy control’s fibroblasts showed homogeneous staining that appears to colocalize with the endoplasmic reticulum marker calnexin at the perinuclear space. The nucleus marker DAPI is shown in blue. Scale bar is 50 μm. (B) ELOVL4 staining in the patient’s fibroblasts showed mislocalization of the protein beyond the perinuclear region with a punctate and aggregated appearance. DAPI = 4',6-Diamidino-2-phenylindole; SCA34 = spinocerebellar ataxia 34.

underwent a complete skin examination. It is possible that a few patients, along with the early gingivitis episodes in 3 subtle skin lesions may have been overlooked by patients patients, may represent milder mucocutaneous involvement or themselves, but this is unlikely to explain alone the absence of coincidental findings. In the previously reported patients with EKV in our patients because the cutaneous findings reported in the same mutation, skin disease was shown to have incomplete previous studies were pronounced and symptomatic.3,6 Ab- penetrance,3 and absence of EKV in SCA34 families has been sence of EKV may reflect the role of modifier genes or envi- reported in patients of Japanese origin.4 ronmental factors involved in cutaneous disease or possibly a milder disease severity that would have gone unnoticed. The Immunofluorescence analyses using dermal fibroblasts from late-onset skin dryness and nummular dermatitis reported by an affected patient revealed mislocalization and aggregation

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 of the ELOVL4 protein. Previous studies did not demon- syndrome because the cognitive deficits are also a direct strate significant serum lipid anomalies in patients with consequence of cerebellar degeneration. Similarly, the skin SCA34, and cellular abnormalities in human tissue had not biopsy findings would have to be confirmed in tissue speci- been reported.3,4 ELOVL4 is involved in the elongation of mens from additional patients with different mutations to VLCFA with carbon chains of 24 atoms or longer and confirm that the dominant negative effect on protein local- catalyzes the rate-limiting reaction in the production of 2 ization observed here is a ubiquitous mechanism in this distinct products: very-long-chain saturated fatty acids disorder. (VLC-SFAs), which are involved in skin barrier formation and sphingolipid production, and VLC polyunsaturated fatty Our findings support the role of ELOVL4 in cerebellar acids, which are essential for photoreceptor cells of the function and present a more precise and exhaustive charac- macula and also for production of sphingolipids.24 In the terization of the SCA34 phenotype. Clinicians should con- cerebellum, the protein is located in all parts of the cerebellar sider the diagnosis of SCA34 (ATX-ELOVL4) even in the cortex and deep cerebellar nuclei, and it is mainly expressed absence of EKV and systematically assess cognitive and in neurons and oligodendrocytes,25 which is compatible with psychiatric features with specific emphasis on executive, the role of VLC-SFA in myelin production. Skin appears as visuospatial, and attention deficits. Analysis of dermal a relevant tissue to investigate the biochemical anomalies fibroblasts from a patient with SCA34 supports a dominant associated with ELOVL4 mutations, considering that the negative effect on ELOVL4 localization. More research is brain and skin are the 2 human organs with the highest needed to understand the specificroleofELOVL4and concentrations of sphingolipids, some of which are de- VLCFA in the cerebellum to better understand the specific pendent on the presence of ELOVL4 for synthesis.26 The pathologic mechanisms of selective cerebellar dysfunction in mislocalization and aggregation of the ELOVL4 protein SCA34. observed in the patient’s fibroblast cells appear similar to those found in previous studies of Stargardt-like macular Acknowledgment dystrophy type 3 (STGD3), which is associated with The authors thank Bastien Par´e for his contribution to the – ELOVL4 dominant mutations.27 29 More specifically, Logan fibroblast cell isolation and culture. Affiliation: Laval et al.27 demonstrated that ELOVL4 mutations associated University Experimental Organogenesis Research Center/ with STGD3 cause mislocalization of the protein beyond the LOEX, Division of Regenerative Medicine, CHU de Qu´ebec endoplasmic reticulum in a punctate and aggregated ap- Research Center – Enfant-J´esus Hospital, Qu´ebec, Canada. pearance very similar to what we observed. In this study, the mutant protein was shown to interact with wild-type Study funding ELOVL4 to alter the localization of the protein and to ex- No targeted funding reported. ertadominantnegativeeffect on enzymatic activity that worsened if the protein was redirected to the endoplasmic Disclosure reticulum.27 Therefore, it is likely that the c.504G>C Disclosures available: Neurology.org/NG. ELOVL4 mutation may also exert its pathogenic effect through a dominant negative mechanism that involves in- Publication history teraction with wild-type ELOVL4, mislocalization, and Received by Neurology: Genetics September 26, 2019. Accepted in final possibly reduction in the production of VLCFA. Although form January 21, 2020. the skin biopsy was performed in the patient that reported cutaneous involvement, we expect that the ELOVL4 mis- ’ localization and aggregation observed in this patient sskin Appendix Authors fibroblasts would be similar in other patients carrying the same mutation and underlies the pathophysiologic mecha- Name Location Role Contribution nism of the disorder in other tissues as well, including the Marie CHU de Qu´ebec- Author Evaluated patients, cerebellum. Beaudin, Universit´e Laval collected the data, MD, MSc analyzed the data, and drafted the manuscript As expected in the evaluation of a rare disease, this study is Leila CHU de Qu´ebec- Author Performed cognitive limited by the small number of patients who originated from Sellami, Universit´e Laval evaluations, analyzed a single family such that the clinical features described here MD, MSc the data, and revised the fl manuscript for may not re ect the whole range of clinical involvement in intellectual content patients with different mutations. This was highlighted by fi Christian CHU de Qu´ebec Author Performed fibroblast the comparison between the ndings from the present and Martel, MSc Research Center isolation and culture, previous studies (table e-1, links.lww.com/NXG/A228), performed fi immunofluorescence, which showed the broad range of clinical ndings in patients and revised the with SCA34. Nevertheless, it is likely that the neurocognitive manuscript for fi intellectual impairment described here is not mutation-speci cand content would be present in all patients with a cerebellar motor

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 3. Cadieux-Dion M, Turcotte-Gauthier M, Noreau A, et al. Expanding the clinical Appendix (continued) phenotype associated with ELOVL4 mutation: study of a large French-Canadian family with autosomal dominant spinocerebellar ataxia and erythrokeratodermia. JAMA Neurol 2014;71:470–475. Name Location Role Contribution 4. Ozaki K, Doi H, Mitsui J, et al. A novel mutation in ELOVL4 leading to spinocer- ebellar ataxia (SCA) with the hot cross bun sign but lacking erythrokeratodermia: ´ Lydia CHU de Quebec Author Performed fibroblast a broadened spectrum of SCA34. JAMA Neurol 2015;72:797–805. Touzel- Research Center isolation and culture, fi ˆ 5. Bourassa CV, Raskin S, Sera ni S, Teive HA, Dion PA, Rouleau GA. A new ELOVL4 Deschenes, performed mutation in a case of spinocerebellar ataxia with erythrokeratodermia. JAMA Neurol MSc immunofluorescence, 2015;72:942–943. and revised the 6. Bourque PR, Warman-Chardon J, Lelli DA, et al. Novel ELOVL4 mutation associated manuscript for with erythrokeratodermia and spinocerebellar ataxia (SCA 34). Neurol Genet 2018;4: intellectual content e263. 7. Schmahmann JD, Sherman JC. The cerebellar cognitive affective syndrome. Brain Gabrielle Montreal Author Performed genetic data 1998;121(pt 4):561–579. Houle, BSc Neurological analysis and 8. Manto M, Marien P. Schmahmann’s syndrome—identification of the third corner- Institute – McGill contributed to drafting stone of clinical ataxiology. Cerebellum Ataxias 2015;2:2. University the manuscript for 9. Laforce R Jr, Buteau JP, Bouchard JP, Rouleau GA, Bouchard RW, Dupre N. Cog- genetics section nitive impairment in ARCA-1, a newly discovered pure cerebellar ataxia syndrome. Cerebellum 2010;9:443–453. Laurence CHU de Qu´ebec- Author Evaluated patients and 10. Pedroso JL, Franca MC Jr, Braga-Neto P, et al. Nonmotor and extracerebellar features Martineau, Universit´e Laval revised the manuscript in Machado-Joseph disease: a review. Mov Disord 2013;28:1200–1208. MD for intellectual content 11. Fancellu R, Paridi D, Tomasello C, et al. Longitudinal study of cognitive and psy- chiatric functions in spinocerebellar ataxia types 1 and 2. J Neurol 2013;260: Kevin CIUSS de la Author Evaluated patients and 3134–3143. Lacroix, MD Mauricie-et-du- revised the manuscript 12. Nasreddine ZS, Phillips NA, Bedirian V, et al. The Montreal Cognitive Assessment, Centre-du-Qu´ebec, for intellectual content MoCA: a brief screening tool for mild cognitive impairment. J Am Geriatr Soc 2005; Trois-Rivi`eres 53:695–699. 13. McKenna A, Hanna M, Banks E, et al. The Genome Analysis Toolkit: a MapReduce Andr´eane CHU de Qu´ebec- Author Evaluated patients and framework for analyzing next-generation DNA sequencing data. Genome Res 2010; Lavall´ee, Universit´e Laval revised the manuscript 20:1297–1303. MD for intellectual content 14. Wang K, Li M, Hakonarson H. ANNOVAR: functional annotation of genetic variants from high-throughput sequencing data. Nucleic Acids Res 2010;38:e164. Nicolas Centre M`ere- Author Evaluated patients and 15. Pare B, Touzel-Deschenes L, Lamontagne R, et al. Early detection of structural Chrestian, Enfant-Soleil – revised the manuscript abnormalities and cytoplasmic accumulation of TDP-43 in tissue-engineered skins MD Universit´e Laval for intellectual content derived from ALS patients. Acta Neuropathol Commun 2015;3:5. 16. Adzhubei IA, Schmidt S, Peshkin L, et al. A method and server for predicting dam- Guy A. Montreal Author Contributed to study aging missense mutations. Nat Methods 2010;7:248–249. Rouleau, Neurological design, supervised 17. Manichaikul A, Mychaleckyj JC, Rich SS, Daly K, Sale M, Chen WM. Robust re- MD, PhD Institute – McGill genetic data analysis, lationship inference in genome-wide association studies. Bioinformatics 2010;26: University and revised the 2867–2873. manuscript for 18. Kansal K, Yang Z, Fishman AM, et al. Structural cerebellar correlates of cog- intellectual content nitive and motor dysfunctions in cerebellar degeneration. Brain 2017;140: – François CHU de Qu´ebec Author Contributed to study 707 720. Gros-Louis, Research Center design, contributed to 19. Stoodley CJ, Valera EM, Schmahmann JD. Functional topography of the cere- bellum for motor and cognitive tasks: an fMRI study. Neuroimage 2012;59: PhD drafting the manuscript – for 1560 1570. immunofluorescence 20. Stoodley CJ, Schmahmann JD. Evidence for topographic organization in the cere- ff section, and revised the bellum of motor control versus cognitive and a ective processing. Cortex 2010;46: – manuscript for 831 844. intellectual content 21. Salmi J, Pallesen KJ, Neuvonen T, et al. Cognitive and motor loops of the human cerebro-cerebellar system. J Cogn Neurosci 2010;22:2663–2676. – Robert CHU de Qu´ebec- Author Performed cognitive 22. Carrera E, Tononi G. Diaschisis: past, present, future. Brain 2014;137:2408 2422. Laforce Jr, Universit´e Laval evaluations, contributed 23. Morihara R, Yamashita T, Deguchi K, et al. Familial and sporadic chronic progressive – MD, PhD to study design, and degenerative parietal ataxia. J Neurol Sci 2018;387:70 74. ff ff revised the manuscript 24. Agbaga MP. Di erent mutations in ELOVL4 a ect very long chain fatty acid bio- for intellectual content synthesis to cause variable neurological disorders in humans. Adv Exp Med Biol 2016; 854:129–135. Nicolas CHU de Qu´ebec- Author Designed and 25. Sherry DM, Hopiavuori BR, Stiles MA, et al. Distribution of ELOVL4 in the de- Dupr´e, MD, Universit´e Laval conceptualized the veloping and adult mouse brain. Front Neuroanat 2017;11:38. MSc study, evaluated 26. Brush RS, Tran JT, Henry KR, McClellan ME, Elliott MH, Mandal MN. Retinal patients, and revised the sphingolipids and their very-long-chain fatty acid-containing species. Invest Oph- – manuscript for thalmol Vis Sci 2010;51:4422 4431. intellectual content 27. Logan S, Agbaga MP, Chan MD, et al. Deciphering mutant ELOVL4 activity in autosomal-dominant Stargardt macular dystrophy. Proc Natl Acad Sci U S A 2013; 110:5446–5451. 28. Grayson C, Molday RS. Dominant negative mechanism underlies autosomal domi- References nant Stargardt-like macular dystrophy linked to mutations in ELOVL4. J Biol Chem 1. Rossi M, Anheim M, Durr A, et al. The genetic nomenclature of recessive cerebellar 2005;280:32521–32530. ataxias. Mov Disord 2018;33:1056–1076. 29. Maugeri A, Meire F, Hoyng CB, et al. A novel mutation in the ELOVL4 gene causes 2. Giroux JM, Barbeau A. Erythrokeratodermia with ataxia. Arch Dermatol 1972;106: autosomal dominant Stargardt-like macular dystrophy. Invest Ophthalmol Vis Sci 183–188. 2004;45:4263–4267.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 ARTICLE OPEN ACCESS Use of local genetic ancestry to assess TOMM40-5239 and risk for Alzheimer disease

Parker L. Bussies, BS, Farid Rajabli, PhD, Anthony Griswold, PhD, Daniel A. Dorfsman, BA, Correspondence Patrice Whitehead, BS, Larry D. Adams, BA, Pedro R. Mena, MD, Michael Cuccaro, PhD, Dr. Vance [email protected] Jonathan L. Haines, PhD, Goldie S. Byrd, PhD, Gary W. Beecham, PhD, Margaret A. Pericak-Vance, PhD, or Dr. Young Juan I. Young, PhD, and Jeffery M. Vance, MD, PhD [email protected] Neurol Genet 2020;6:e404. doi:10.1212/NXG.0000000000000404 Abstract Objective Here, we re-examine TOMM40-5239 as a race/ethnicity-specific risk modifier for late-onset Alzheimer disease (LOAD) with adjustment for local genomic ancestry (LGA) in Apolipo- protein E (APOE) «4 haplotypes.

Methods The TOMM40-5239 size was determined by fragment analysis and whole genome sequencing in homozygous APOE «3 and APOE «4 haplotypes of African (AF) or European (EUR) ancestry. The risk for LOAD was assessed within groups by allele size.

Results The TOMM40-5239 length did not modify risk for LOAD in APOE «4 haplotypes with EUR or AF LGA. Increasing length of TOMM40-5239 was associated with a significantly reduced risk for LOAD in EUR APOE e3 haplotypes.

Conclusions Adjustment for LGA confirms that TOMM40-5239 cannot explain the strong differential risk for LOAD between APOE e4 with EUR and AF LGA. Our study does confirm previous reports that increasing allele length of the TOMM40-5239 repeat is associated with decreased risk for LOAD in carriers of homozygous APOE e3 alleles and demonstrates that this effect is occurring in those individuals with the EUR LGA APOE e3 allele haplotype.

From the John P. Hussman Institute for Human Genomics (P.L.B., F.R., A.G., D.A.D., P.W., L.D.A., P.R.M., M.C., G.W.B., M.A.P.-V., J.I.Y., J.M.V.), Miller School of Medicine, University of Miami; Dr. John T. MacDonald Foundation Department of Human Genetics (A.G., M.C., G.W.B., M.A.P.-V., J.I.Y., J.M.V.), Miller School of Medicine, University of Miami; Department of Population and Quantitative Health Sciences (J.L.H.), Institute for Computational Biology, Case Western Reserve University School of Medicine, Cleveland, OH; and Wake Forest School of Medicine (G.S.B.), Bowman Gray Center for Medical Education, Winston-Salem, NC.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by NIH grant.

Previous abstract presentation: The TOMM40 Repeat Does Not Provide Protection for AD for APOE4 Carriers with Local Genomic African Ancestry (LGA); AAIC 2019; Los Angeles, CA. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary AF = African; APOE = Apolipoprotein E; EUR = European; IRB = institutional review board; LOAD = late-onset Alzheimer disease; LGA = local genomic ancestry; SNP = single nucleotide polymorphism; WGS = whole genome sequencing.

Apolipoprotein E (APOE) has long been the strongest and most were obtained for all participants. This included de- consistently identified gene associated with the risk for de- mographic variables, diagnosis, age at onset for cases, veloping late-onset Alzheimer disease (LOAD).1,2 Of interest, method of diagnosis, history of illness, and the presence of the APOE «4 allele in AFs and African Americans confers other relevant comorbidities. The Clinical Dementia Rating a much lower risk for disease than the identical allele in Scale,20,21 National Institute on Aging (NIA)-Late Onset – Europeans (EUR).3 5 It was recently discovered that this racial Alzheimer’s Disease cognitive battery,22,23 and Mini-Mental difference can be attributed to the local genetic ancestry (LGA) State Exam24 or the Modified Mini-Mental State Exam25 within the APOE haplotype.6 The responsible “protective” lo- readings were collected for all participants. cus within the APOE LGA remains unknown, however. TOMM40 is a gene lying within the APOE haplotype that Diagnostic criteria codes for a channel-forming subunit required for protein im- Diagnosis was made via case conferences of all clinical, historical, port into mitochondria. It contains an intronic poly-T repeat and psychometric test data (e.g., laboratory tests, neurologic known as “TOMM40-5239,” which varies in length by in- 9 examination, neuroimaging, and neuropsychological screening dividual and race/ethnicity. The TOMM40-523 length has and testing) by a multidisciplinary clinical adjudication panel been proposed to influence the transcription of APOE and thus – consisting of a boarded neurologist (J.M.V.), neuropsychologist modify the risk for LOAD,7 13 although the significance of its – (M.L.C.), and clinical staff (P.R.M.). All individuals with AD association with LOAD remains controversial.14 19 The 9 met the internationally recognized standard National Institute TOMM40-523 LOAD relationship has been analyzed by race of Neurological Disorders and Stroke - Alzheimer’sDiseaseand in the context of global genetic ancestry but not by using ad- Related Disorders Association26 or updated NIA-Alzheimer’s justment for LGA within the TOMM40-APOE haplotype, fi fi fi Association criteria and were further classi ed as de nite allowing for misclassi cation, given sometimes common dif- (neuropathologic confirmation), probable, or possible AD.27 ferences between LGA and global ancestry.6 fi ff fi 9 Individuals classi ed as una ected controls were older than 65 To con rm whether varying sizes of the TOMM40-523 allele years of age, cognitively normal on the NIA-Late Onset can truly explain the reduced risk for LOAD in African ’ « Alzheimer s Disease battery, and had a CDR of zero. The Americans expressing the APOE 4 allele, we re-examine controls were matched to cases for age, sex, and ethnicity. TOMM40-5239 with adjustment for LGA. We further in- vestigate racial differences in effect of TOMM40-5239 on « APOE genotyping and determination of local genetic ancestry APOE 3 carriers. genome-wide single nucleotide polymorphism (SNP) geno- typing was performed using 3 different platforms: Expanded Multi-Ethnic Genotyping Array, Illumina 1Mduo (v3), and the Methods Global Screening Array (Illumina, San Diego, CA). APOE 28 Standard protocol approvals, registrations, genotyping was conducted as in Saunders et al. Quality con- 29 and patient consents trol analyses were executed using the PLINK software, v.2. All ascertainment activities were approved by the institutional The samples with call rates less than 90% and with excess or review boards (IRBs) at the respective universities and ad- insufficient heterozygosity (±3 SDs) were excluded. Sex con- hered to the tenets of the Declaration of Helsinki. Informed cordance was checked using X chromosome data. To eliminate written consent was obtained for all participants. The current duplicate and related samples, relatedness among the samples study is approved by the University of Miami IRB. was estimated using identity by descent. SNPs were eliminated from further analysis if they were present in samples with call Samples and ascertainment rates less than 97%, had minor allele frequencies less than 0.01, − DNA samples used for this study were part of a larger study or were not in Hardy-Weinberg equilibrium (p < 1.e 5).30 of Alzheimer disease (AD) genetics and were ascertained at Case Western University, Wake Forest University, and the Genotyping data were initially phased using the SHAPEIT University of Miami between 2012 and 2019. Participants tool ver.2 to identify local ancestry at the TOMM40-APOE were selected if they were homozygous for either the APOE haplotype.31 The discriminative RFMix modeling approach «3orAPOE «4 allele, eliminating the need to phase was used to estimate the genetic ancestral background (AF vs TOMM40-5239 alleles by haplotype. All participants were EUR) at the region of interest.32 The Human Genome Di- enrolled under a standard ascertainment protocol. As part of versity Project (HGDP) data were used as the reference panel this protocol, sociodemographic and clinical historic data in the analysis.33

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Genotyping of TOMM40-5239 allele using long, short vs very long, and short vs sum of long and very fragment analysis long). The TOMM40-5239 poly-T repeat was first PCR-amplified using fluorescein amidite-labeled forward primer 59– Data availability CCTCCAAAGCATTGGGATTA-39 and reverse primer 59- Any data pertaining to this article and not published within GATTCCTCACAACCCCAAGA-39. PCR was performed this article is publicly available and may be requested through using Taq polymerase in the presence of 4% DMSO with collaboration. a final volume of 25 uL. PCR conditions were as follows: initial 6-minute denaturation at 94°C, followed by 30 cycles of denaturation at 94°C for 30 seconds, annealing at 54°C for 30 Results seconds, and extension at 72°C for 40 seconds. Final exten- Descriptive statistics sion was performed for 10 minutes at 72°C. PCR products A total of 205 samples composed of 75 controls and 130 cases underwent subsequent fragment analysis with LIZ600 size were analyzed, making up 410 individual haplotypes. Within standard using 3130xl Genetic Analyzer, and resolved frag- AF LGA samples, 48 were APOE «3/«3 (15 cases, 33 con- ments were visualized with GeneMapper v4.0 (Applied Bio- trols) and 47 were APOE «4/«4 (38 cases, 9 controls). Within systems Inc., CA). EUR LGA samples, 47 were APOE «3/«3 (34 cases, 13 controls) and 63 were APOE «4/«4 (43 cases, 20 controls) The allele size was determined by subtracting the fragment (table 1). analysis-determined size by 439, the number of base pairs surrounding each end of the poly-T stretch in our PCR The average age at onset for affected individuals was 71.3 product. The mode of each peak was selected and established years old (SD = 8.1 years). The average age of examination for “ ” fi as the true poly-T length ( gure e-1, links.lww.com/NXG/ controls was 73.0 year old (SD = 6.9 years). Affected and A230). In cases where the mode was shared by 2 peaks, the control groups consisted of 70.8% and 65.3% women, re- mean length of the 2 peaks was used. In all cases, peak size was spectively. The entirety of each haplotype was either EUR rounded to the nearest integer. A replicate fragment analysis LGA or AF LGA, with zero samples expressing mixed LGA. was performed on 20 randomly selected samples to ensure Within affected individuals, 56.0% of haplotypes exhibited AF fl consistency in length calls. A uorescein amidite-labeled PCR LGA. 40.8% of control haplotypes contained AF LGA. fragment with known length and void of repeated sequences was run alongside samples to ensure accurate sizing. Genotyping Polymerase stutter for the TOMM40-5239 allele, as de- Genotyping of TOMM40-5239 using Whole scribed in previous TOMM40 studies,18 was observed with Genome Sequencing an average of 4 peaks per allele (figure e-1, links.lww.com/ To test whether whole genome sequencing (WGS) could NXG/A230). There was 100% congruence of peaks between be used to genotype the TOMM40-5239 repeat, WGS was replicate samples. The length standard was consistently performed on a subset of 15 samples using a PCR-free found to be 4bp shorter than its known value. To correct for library preparation and paired-end 150bp sequencing on this discrepancy, 4bp were added to poly-T allele lengths of the Illumina Novaseq. Raw reads were aligned to the human each sample. Comparable with previous study,36 our allele reference genome GRCh38 using the Burrows-Wheeler sizes appeared to distribute within 4 separate bins rather transform alignment algorithm.34 Resulting Binary Align- than the traditionally used 3 bin sizes (figure 1). For com- ment Map files were visualized in the Integrative Genomics pleteness, we also performed a similar 4-bin analysis in which Viewer,35 and the number of thymine bases at rs10524523 the allele sizes were binned into very small (≤22 T’s), small were calculated and compared with fragment analysis of the (23–28 T’s), long (29–34 T’s), and very long (≥35 T’s). same set of samples. Allele comparisons made can be seen in table e-1, links.lww. com/NXG/A231. Statistical analysis TOMM40-5239 poly-T lengths were firstbinnedintoshort (<20 T’s), long (20–29T’s), and very long (≥30 T’s) alleles, as originally defined.18 To assess the effect of the TOMM40-5239 Table 1 Descriptive statistics poly-T length on LOAD risk, we stratified our data set AF LGA EUR LGA according to local ancestries (AF and EUR) and APOE alleles APOE «3/«3 15/33 34/13 (e3 and e4). Next, we categorized the counts of individuals in our data set by length of TOMM40-5239 poly-T (short, long, APOE «4/«4 38/9 43/20 ff ff and very long) and a ection status (a ected vs control). We Abbreviations: AF = African; APOE = Apolipoprotein E; EUR = European; LGA = applied the Fisher exact test on each subgroup to test the local genetic ancestry. fi ff Number of case (unbold) and control (bold) samples analyzed within each signi cance of a di erence between the proportions in the LGA (AF and EUR) and APOE genotype (e3/e3 and e4/e4). length of TOMM40-5239 poly-T and affection status (short vs

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Figure 1 Distribution of TOMM40-5239 allele sizes across all samples

Allele count (y-axis) assigned to respective poly-T allele lengths (x-axis) for both cases (dark red) and controls (light red). The dotted lines delineate the 4 bin sizes used in the four bin analysis. LOAD = late-onset Alzheimer disease.

Allele frequencies differ by ancestry and exhibited the most variance in allele size, and EUR APOE «4 APOE status contained more long (L) alleles than any other haplotype. Allele frequencies varied between haplotypes harboring EUR vs AF ancestry and APOE «3vsAPOE «4 alleles (figure 2). AF Effect of TOMM40-5239 length on risk for LOAD LGA was more than 2 times more likely to display short (S) There was no significant difference in risk for LOAD between alleles compared with EUR LGA (148 vs 70). Long (L) alleles “S” and “L” alleles, “S” and “very long (VL)” alleles, or “S” were commonly observed in APOE «4 (123 alleles) but not and “L+VL” alleles in the APOE «4 haplotypes with AF (p = APOE «3 (24 alleles) haplotypes. AF APOE «4 haplotypes 0.2269, 0.2055, 0.7957, respectively) or EUR (p = 1, 0.9954, 1,

Figure 2 TOMM40-5239 allele frequencies

TOMM40-5239 allele frequencies in AF APOE e3(A),AFAPOEe4 (C), EUR APOE e3 (B), and AF APOE e3 (D) haplotypes for both cases (dark red) and controls (light red). AF = African; APOE = Apoli- poprotein E; EUR = European; L = long; LGA = local genetic ancestry; LOAD = late-onset Alzheimer disease; S = short; VL = very long.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 2 Whole Genome Sequencing vs Fragment Analysis

Sample FA Allele 1 WGS Allele 1 FA Allele 2 WGS Allele 2

201616933 15 14 32 32

201803389 16 15 16 15

201600975 16 15 16 15

201802895 15 14 28 28

201609823 21 21 34 30

201609813 16 15 33 14

201609811 16 15 32 31

201600974 16 15 16 13

201710863 16 15 34 15

201706978 15 14 38 13

201600962 16 15 16 15

201706996 16 15 21 21

201802891 16 15 28 15

201802829 16 15 28 28

201802899 16 15 28 28

Abbreviations: FA = fragment analysis; WGS = whole genome sequencing. Comparison of TOMM40-5239 genotyping results between WGS and FA. Allele sizes are listed as number of T’s. respectively) LGA. However, there was a significant effect of larger sample size is needed to test any possible weak to mod- TOMM40-5239 allele size within APOE «3 haplotypes of erate significant differences, although given the current p-values, EUR LGA. Both “VL” and “VL + L” allele groups had reduced this seems unlikely. Therefore, we suggest that future efforts to risk compared with haplotypes with an “S” allele (VL: explain the large risk difference between AFs and non-Hispanic p = 0.0104) (VL + L: p = 0.011) (table 2). Our 4-bin analysis whites should be directed toward the investigation of other (VS, S, L, and VL) results were congruent with those of our candidate regions within the APOE haplotype to identify the 3-bin analysis (table e-1, links.lww.com/NXG/A231). protective variant for APOE «4foundintheAFLGA.

Genotyping TOMM40-5239 with WGS Of interest, we did find a significant relationship between the Fifteen samples were genotyped using WGS. Their allele poly-T length and risk for LOAD in APOE «3 alleles sizes, compared with the allele sizes determined via fragment harboring EUR LGA. The increasing number of T’s was as- analysis, can be seen in table 3. The correlations were calcu- sociated with decreased risk for LOAD. Our findings validate lated using fragment analysis as the reference length. The 2 the findings of 2 previous studies reporting a significant methods correlated well for alleles <20 bp in length under-representation of VL-APOE «3 haplotypes in the 2 (R = 0.95), though correlation decreased with inclusion of LOAD cases vs controls36 and significantly fewer VL alleles in 2 2 “L” (R = 0.85) and “VL” (R = 0.57) allele sizes. EUR LOAD patients vs controls.37 This suggests that the TOMM40-5239 variant may indeed play a role in modifying Discussion risk for disease in this subpopulation of APOE «3 carriers. Given the tendency for human genomic repeats to have This TOMM40-5239 LOAD association study adjusted for local a negative phenotypic effect as they increase in length,38,39 the rather than global genetic ancestry, eliminating the potential for finding that the largest expansion of the TOMM40-5239 re- misclassification of ancestry housed within the TOMM40-5239 peat is protective is unusual at first glance. Our results align locus. This is an important distinction which cannot be over- biologically in the case of the TOMM40-5239 repeat; how- looked. After re-examination of the TOMM40-5239 allele with ever, as studies have shown that the expression of TOMM40 – adjustment for LGA, our study found no significant effect of the increases with size of TOMM40-5239 poly-T repeat11 13 and TOMM40-5239 poly-T size on the risk for LOAD within APOE that augmented TOMM40 expression confers subsequent «4 haplotypes. This suggests that the TOMM40-5239 allele mitochondrial protection.11 TOMM40-5239 has additionally cannot explain the strong difference in risk conferred by the been identified as a transcriptional start site by the FANTOM APOE «4 allele between AF and EUR carriers of APOE «4. A project.40 It is also possible, and perhaps likely, that the

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Table 3 The effect of TOMM40-5239 allele size on the risk for LOAD

Haplotype Comparison OR CI p Value

AF «4 S vs L 1.53 0.3–7.9 0.2269

S vs VL 0.43 0.1–1.7 0.2055

S vs (L + VL) 0.83 0.2–2.7 0.7957

EUR «4 S vs L 0.78 0–10.1 1

SvsVL 0.30 0–5.5 0.5594

S vs (L + VL) 0.71 0–9.2 1

AF «3 SvsL —— —

S vs VL 1.49 0.4–4.8 0.5742

S vs (L + VL) 1.36 0.4–4.4 0.5873

EUR «3 SvsL —— —

S vs VL 0.27 0.1–0.8 0.0104

S vs (L + VL) 0.28 0.1–0.8 0.0110

Abbreviations: AF = African; APOE = Apolipoprotein E; EUR = European; L = long; LGA = local genetic ancestry; S = short; VL = very long. Statistical test results between TOMM40-5239 allele sizes among APOE e3 and APOE e4 alleles in the presence of AF and EUR LGA. “SvsL” could not be calculated because of insufficient “L” alleles.

TOMM40-5239 repeat is a tagging variant for a distinct pro- In our study, the TOMM40-5239 poly-T variant could not tective factor which is APOE «3 specific. Stratifying risk within explain the large difference in the risk for LOAD between AF patients carrying APOE «3onaEURLGA,mostnon-Hispanic and EUR LGA on the APOE e4 haplotype. Increasing white patients, could be important in the clinical setting and TOMM40-5239 length on the APOE e3 haplotype may de- perhaps in clinical trials. Further molecular experiments in in- crease the risk for LOAD and warrants further investigation. ducible pluripotent stem cell lines with APOE e3/e3 possessing The differential effect of LGA in APOE «4, and now shown in EUR LGA could deepen our understanding of this association APOE «3, is continuing the examples of the importance of and provide clues regarding LOAD pathophysiology. considering LGA while interpreting the risk for disease. In the setting of personalized medicine, identical genetic variants do Our study further used WGS to genotype the TOMM40- not always confer equal risk. Rather, genetic loci must be 5239 allele. Genotyping long microsatellite regions such as examined with an understanding of their contextual ancestry. TOMM40-5239 can be challenging. Current methods used As we increase the amount of diversity in our genetic studies, to size poly-T and poly-A tracts such as TOMM40-5239 this is likely to become a common cofactor in assessing risk. within the human genome include fragment analysis and Sanger sequencing.14 Although our WGS data correlated Acknowledgment extremely well with “S” TOMM40-5239 alleles, it grew less The authors thank the families and community members who consistent with the inclusion of “L” and “VL” alleles. This is graciously agreed to participate in the study and made this undoubtedly because of the shorter size fragments used in research possible. Illumina WGS. Long fragment WGS would likely increase the concordance to fragment analysis. Repetitive DNA Study funding regions are likely to gain importance in examining the risk for This work was supported by the NIH [RF1-AG059018, R01- disease and use of WGS to assess these regions and may play AG028786, RF1-AG054074, U01-AG052410], the Alz- a progressively large role. heimer’s Association [Zenith Award, ZEN-19-591586], and the BrightFocus Foundation [A2018425S]. As noted, there were very few “L” alleles within the APOE «3 haplotypes. Because of this, we were unable to generate Disclosure proper odds ratios for “S” vs “L” alleles in this subgroup. It was The authors have no competing interests to declare. Go to additionally difficult to obtain the AF LGA APOE «4/«4 Neurology.org/NG for full disclosures. control samples. Although we improved on previous study designs by adjusting for LGA, our reduced sample size, as Publication history 16,36 compared to previous studies, may have limited the res- Received by Neurology: Genetics September 1, 2019. Accepted in final olution within groups. form January 14, 2020.

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 10. NishimuraA,NonomuraH,TanakaS,et al. Characterization of APOE and Appendix Authors TOMM40 allele frequencies in the Japanese population. Alzheimer Dement 2017; 3:524–530. 11. Zeitlow K, Charlambous L, Ng I, et al. The biological foundation of the genetic Name Location Contribution association of TOMM40 with late-onset Alzheimer’s disease. Biochim Biophys Acta Mol Basis Dis 2017;1863:2973–2986. Parker L. University Conducted the experiment, analyzed 12. Payton A, Sindrewicz P, Pessoa V, et al. A TOMM40 poly-T variant modulates gene Bussies, BS of Miami and interpreted the data, and wrote expression and is associated with vocabulary ability and decline in nonpathologic the manuscript aging. Neurobiol Aging 2016;39:217.e1–217.e7. 13. Linnertz C, Anderson L, Gottschalk W, et al. The cis-regulatory effect of an Alz- Farid Rajabli, University Interpreted the data and wrote the heimer’s disease-associated poly-T locus on expression of TOMM40 and apolipo- PhD of Miami manuscript protein E genes. Alzheimer’s Demen 2014;10:541–551. 14. Chiba-Falek O, Gottschalk WK, Lutz MW. The effects of the TOMM40 poly-T alleles Anthony University Interpreted the data and wrote the on Alzheimer’s disease phenotypes. Alzheimer’s Demen 2018;14:692–698. Griswold, PhD of Miami manuscript 15. Roses AD, Lutz MW, Crenshaw DG, Grossman I, Saunders AM, Gottschalk WK. TOMM40 and APOE: requirements for replication studies of association with age Daniel A. University Prepared the samples, maintained of disease onset and enrichment of a clinical trial. Alzheimer’s Demen 2013;9: Dorfsman, BA of Miami clinical database, and oversaw 132–136. regulatory compliance 16. Jun G, Vardarajan BN, Buros J, et al. Comprehensive search for Alzheimer disease susceptibility loci in the APOE region. Arch Neurol 2012;69:1270–1279. Patrice University Prepared the samples, maintained 17. Lutz MW, Crenshaw DG, Saunders AM, Roses AD. Genetic variation at a single locus Whitehead, BS of Miami clinical database, and oversaw and age of onset for Alzheimer’s disease. Alzheimer’s Demen 2010;6:122–131. regulatory compliance 18. Roses A, Lutz MW, Amrine-Madsen H, et al. A TOMM40 variable-length poly- morphism predicts the age of late-onset Alzheimer’s disease. Pharmacogenomics J Larry D. Adams, University Designed and conducted 2010;10:375. BA of Miami ascertainment process 19. Caselli RJ, Saunders A, Lutz M, Huentelman M, Reiman E, Roses A. TOMM40, APOE, and age of onset of Alzheimer’s disease. Alzheimer Dement 2010;6:S202. Pedro R. Mena, University Ascertained patients and adjudicated 20. Hughes CP, Berg L, Danziger W, Coben LA, Martin RL. A new clinical scale for the MD of Miami disease status staging of dementia. Br J Psychiatry 1982;140:566–572. 21. Morris JC. The Clinical Dementia Rating (CDR): current version and scoring rules. Michael University Ascertained patients and adjudicated Neurology 1993;43:2412–2414. Cuccaro, PhD of Miami disease status 22. Morris JC, Weintraub S, Chui HC, et al. The uniform data set (UDS): clinical and cognitive variables and descriptive data from alzheimer disease centers. Alzheimer Dis Jonathan L. Case Designed and conducted Associated Disord 2006;20:210–216. Haines, PhD Western ascertainment process 23. Weintraub S, Salmon D, Mercaldo N, et al. The Alzheimer’s disease centers’ uniform University data set (UDS): the neuropsychological test battery. Alzheimer Dis associated Disord 2009;23:91–101. Goldie S. Byrd, Wake Forest Designed and conducted 24. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state”: a practical method for PhD University ascertainment process grading the cognitive state of patients for the clinician. J Psychiatr Res 1975;12: 189–198. Gary W. University Interpreted the data and edited the 25. Teng EL, Chui HC. The Modified Mini-Mental State (3MS) Examination. J Clin Beecham, PhD of Miami manuscript Psychiatry 1987;41:114–121. 26. McKhann G, Drachman D, Folstein M, Katzman R, Price D, Stadlan EM. Clinical Margaret A. University Coordinated collaboration between diagnosis of alzheimer’s disease: report of the NINCDS‐ADRDA work group* under Pericak-Vance, of Miami ascertainment centers, provided the auspices of department of Health and human Services Task Force on alzheimer’s PhD clinical samples, and secured funding disease. Neurology 1984;34:939. 27. McKhann GM, Knopman DS, Chertkow H, et al. The diagnosis of dementia due to Juan I. Young, University Conducted the experiment, analyzed Alzheimer’s disease: recommendations from the National Institute on Aging-Alz- PhD of Miami and interpreted the data, and wrote heimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. the manuscript Alzheimer’s Demen 2011;7:263–269. 28. Saunders AM, Hulette C, Welsh-Bohmer KA, et al. Specificity, sensitivity, and pre- Jeffery M. University Secured funding, designed the dictive value of apolipoprotein-E genotyping for sporadic Alzheimer’s disease. Lancet Vance, MD, PhD of Miami experiment, adjudicated disease 1996;348:90–93. status, wrote the manuscript 29. Purcell S, Neale B, Todd-Brown K, et al. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet 2007;81: 559–575. 30. Anderson CA, Pettersson FH, Clarke GM, Cardon LR, Morris AP, Zondervan KT. Data quality control in genetic case-control association studies. Nat Protoc 2010;5: References 1564. 1. Pericak-Vance MA, Haines JL. Alzheimer’s disease. In: King RA, Rotter JI, eds. The 31. Delaneau O, Marchini J, McVean GA, et al. Integrating sequence and array data to Genetic Basis of Common Diseases. 2nd ed. Oxford, UK: Oxford University Press; create an improved 1000 Genomes Project haplotype reference panel. Nature 2002:818–830. Commun 2014;5:3934. 2. Michaelson DM. Apoe e4: the most prevalent yet understudied risk factor for Alz- 32. Maples BK, Gravel S, Kenny EE, Bustamante CD. RFMix: a discriminative modeling heimer’s disease. Alzheimer’s Demen 2014;10:861–868. approach for rapid and robust local-ancestry inference. Am J Hum Genet 2013;93: 3. Tang MX, Stern Y, Marder K, et al. The APOE-∊ 4 allele and the risk of Alzheimer 278–288. disease among African Americans, Whites, and Hispanics. JAMA 1998;279:751–755. 33. Cann HM, De Toma C, Cazes L, et al. A human genome diversity cell line panel. 4. Mendizabal I, Sandoval K, Berniell-Lee G, et al. Genetic origin, admixture, and Science 2002;296:261–262. asymmetry in maternal and paternal human lineages in Cuba. BMC Evol Biol 2008;8: 34. Li H, Durbin R. Fast and accurate short read alignment with Burrows–Wheeler 213. transform. Bioinformatics 2009;25:1754–1760. 5. Ohara T, Ninomiya T, Hirakawa Y, et al. Association study of susceptibility genes for 35. Robinson JT, Thorvaldsd´ottir H, Winckler W, et al. Integrative genomics viewer. Nat lateonset Alzheimer’s disease in the Japanese population. Psychiatr Genet 2012;22: Biotechnol 2011;29:24–26. 290–293. 36. Maruszak A, Pepło´nska B, Safranow K, Chodakowska-Zebrowska˙ M, Barcikowska M, 6. Rajabli F, Feliciano BE, Celis K, et al. Ancestral origin of APOE e4 Alzheimer disease Zekanowski˙ C. TOMM40 rs10524523 polymorphism’s role in late-onset Alzheimer’s risk in Puerto Rican and African American populations. PLoS Genet 2018;14: disease and in longevity. J Alzheimer’s Dis 2012;28:309–322. e1007791. 37. Cruchaga C, Nowotny P, Kauwe JS, et al. Association and expression analyses with 7. Linnertz C, Saunders AM, Lutz MW, et al. Characterization of the poly-T variant in SNPs in TOMM40 in Alzheimer’s disease. Arch Neurol 2011;68:1013–1019. the TOMM40 gene in diverse populations. PLoS One 2012;7:e30994. 38. Beyer K, Humbert J, Ferrer A, et al. A variable poly‐T sequence modulates α‐ 8. Roses AD, Lutz MW, Saunders AM, et al. African-American TOMM40’523-APOE synuclein isoform expression and is associated with aging. J Neurosci Res 2007;85: haplotypes are admixture of West African and Caucasian alleles. Alzheimer’s Demen 1538–1546. 2014;10:592–601. 39. Sutherland GR, Richards RI. Simple tandem DNA repeats and human genetic disease. 9. Yu L, Lutz MW, Wilson RS, et al. APOE e4-TOMM40 ‘523 haplotypes and the risk of Proc Natl Acad Sci 1995;92:3636–3641. Alzheimer’s disease in older Caucasian and African Americans. PLoS One 2017;12: 40. Bessiere C, Saraswat M, Grapotte M, et al. Sequence-level instructions direct tran- e0180356. scription at polyT short tandem repeats. bioRxiv 2019:634261.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 ARTICLE OPEN ACCESS Genetic testing utilization for patients with neurologic disease and the limitations of claims data

Samuel J. Mackenzie, MD, PhD, Chun Chieh Lin, PhD, MBA, Peter K. Todd, MD, PhD, James F. Burke, MD, MS, Correspondence and Brian C. Callaghan, MD, MS Dr. Callaghan [email protected] Neurol Genet 2020;6:e405. doi:10.1212/NXG.0000000000000405 Abstract Objective To determine the utilization of genetic testing in patients seen by a neurologist within a large private insurance population.

Methods Using the Optum health care claims database, we identified a cross-sectional cohort of patients who had been evaluated by a neurologist no more than 30 days before initial genetic testing. Within this group, we then categorized genetic testing between 2014 and 2016 on the basis of the Current Procedural Terminology (CPT) codes related to molecular and genetic testing. We also evaluated the International Classification of Disease Version 9 Clinical Code Classi- fications (ICD-9 CCS) associated with testing.

Results From 2014 to 2016, a total of 45,014 claims were placed for 29,951 patients who had been evaluated by a neurologist within the preceding 30 days. Of these, 29,926 (66.5%) were associated with codes that were too nonspecific to infer what test was actually performed. Among those claims where the test was clearly identifiable, 7,307 (16.2%) were likely obtained for purposes of neurologic diagnosis, whereas the remainder (17.2%) was obtained for non- neurological purposes. An additional 3,793 claims (8.4%) wherein the test ordered could not be clearly identified were associated with a neurology-related ICD-9 CCS.

Conclusions Accurate assessment of genetic testing utilization using claims data is not possible given the high prevalence of nonspecific codes. Reducing the ambiguity surrounding the CPT codes and the actual testing performed will become even more important as more genetic tests become available.

From the Division of Pediatric Neurology (S.J.M.), Department of Pediatrics, University of Michigan; Department of Neurology (C.C.L, P.K.T., J.F.B, B.C.C.), University of Michigan; Department of Veterans Affairs Ann Arbor Healthcare System (P.K.T., J.F.B, B.C.C.); and the Institute for Healthcare Policy and Innovation (J.F.B, B.C.C.), University of Michigan, Ann Arbor, MI.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary CPT = Current Procedural Terminology; ICD-9 CCS = International Classification of Disease Version 9 Clinical Code Classification.

Clinical genetic testing is an increasingly common component may predispose a patient to cerebrovascular disease were cate- of modern-day neurologic practice, with a rapidly growing set gorized as “neurologic.” of tools available for ascertaining the potential causes of dis- ease. Over the past 2 decades, next-generation sequencing has To better evaluate whether testing was ordered as part of made it possible to probe multiple gene targets on an expo- a neurologic evaluation, we also evaluated codes associated nentially larger scale at relatively low processing costs.1 with neurology-related International Classification of Disease However, this less-targeted approach may identify incidental Version 9 clinical code classifications (ICD-9 CCS). findings and carry additional expenses associated with data interpretation and counseling.2,3 Data queries were performed using SAS 9.4 for Windows (Cary, NC). As more tests come to the market, ordering physicians may struggle with what test or sequence of tests to pursue to Data availability minimize the cost, time to diagnosis, and incidental findings. All data relevant to this study are contained within the article Although most neurologists surveyed in a 2014 study self- and supplemental materials. reported ordering genetic testing on a regular basis, most also reported deficiencies in their knowledge of genetics and ability to interpret tests.4 Ascertaining what tests neurolo- Results gists are commonly ordering in real-world practice would From 2014 to 2016, a total of 45,014 claims were placed for provide a foundation for addressing these problems. Un- initial molecular and genetic testing in 29,951 patients evalu- fortunately, little is known about the utilization of genetic ated by a neurologist within the preceding 30 days. The mean testing on a population-wide scale. patient age was 45 years (SD 20.2 years), and 4,114 patients (10.5%) were younger than the age of 18 years at the time of In this study, we aimed to determine the use of genetic testing claims submission. Most patients (25,162; 84%) had only among patients who were evaluated by an outpatient neu- a single claim associated with testing. Among the remaining rologist from 2014 to 2016 through targeted analysis of a large patients with more than 1 claim submitted as part of their initial set of private insurance claims data. genetic testing, an average of 4.1 ± 2.8 (mean ± SD) claims were placed (range 2–28).

Methods The most common CPT code submitted across all patients Standard protocol approvals, registrations, was 89,240 for “Unlisted miscellaneous pathology test” and patient consent (19,564; 43.5%; figure). The diagnostic test performed This study was approved by the University of Michigan In- could not be ascertained for most codes submitted (29,926; stitutional Review Board. 66.5%; figure). On the basis of testing claims alone, we estimated that approximately 7,307 (16.2%; figure) were Data collection and analysis sent for the purposes of neurologic diagnosis, most com- The Optum health care claims database comprises deiden- monly, to help determine the presence of an underlying tified medical claims data for several million privately insured coagulopathy (CPT 81241, factor V Leiden: 2,115 claims individuals across the country. We identified a cross- [4.7%]; CPT 81291, MTHFR mutation: 1,961 claims sectional cohort of patients who had been evaluated by [4.4%]; CPT 81240, Prothrombin gene mutation: a neurologist between 2014 and 2016. We then extracted 1,773 claims [3.9%]; figure). claims data associated with molecular and genetic testing Current Procedural Terminology (CPT) codes submitted Only 4,823 claims (10.7%) were associated with a neurol- within 30 days of their neurology visit. We categorized the ogy-related ICD-9 CCS. These claims were categorized as tests by diagnostic intent as “neurologic or likely neurologic;” follows: 652 neurologic or likely neurologic, 379 non- “non-neurologic,” or “unknown” on the basis of code neurological, and 3,793 unknown. “Epilepsy; convulsions” description and author consensus (table e-1, links.lww.com/ (745; 1.7%), “Other nervous system symptoms and dis- NXG/A232). All codes related to a molecular diagnosis of orders” (593; 1.3%), “Disorders of the peripheral nervous cancer, including those potentially related to neuro-oncological system” (592; 1.3%), and “Headache; including migraine” diagnoses, were categorized as “non-neurological.” All codes (591; 1.3%), were the most commonly associated classi- related to genetic testing for an underlying coagulopathy that fications. The CPT code 89,240 for the “Unlisted

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure Categorization of molecular and genetic testing CPT codes in patients evaluated by a neurologist within 30 days of testing, 2014–2016

Most codes were categorized as “unknown” on the basis of unknown diagnostic intent and ambiguous test represented (N = 29,926; 66.5%). Among codes categorized as “neurologic or likely neurologic” (N = 7,307, 16.2%), most testing was obtained for diagnosis of underlying coagulopathy. Unabbreviated descriptions of tests are provided in table e-1, links.lww.com/NXG/A232. CPT = Current Procedural Terminology. miscellaneous pathology test” was the most common code with their claims, and this association did not reduce the associated with each neurology-related ICD-9 CCS. ambiguity surrounding what test was ordered.

According to the NIH’s Genetic Testing Registry, at the time of Discussion this writing, there are over 60,000 genetic tests available on the market compared with just 1,038 tests in 2012.5 As the volume Our objective was to determine how genetic testing is being of molecular pathology testing has increased, the American used in patients with neurologic disease on a population- Medical Association’s CPT Molecular Pathology Coding wide scale, although we found that it is extremely difficult to Workgroup transitioned from a methodology-based “stacking evaluate the utilization of genetic testing in a modern claims code” system to new system of “tiered” codes, wherein analyte- database because of the nonspecificity of codes. In our co- specific high-volume tests are given a tier 1 designation and hort of privately insured neurology patients who underwent tests for less common diseases are categorized as tier 2.6,7 Tier 2 genetic testing from 2014 to 2016, most testing was not codes, sometimes referred to as “umbrella” codes, given that identifiable on the basis of CPT coding because two-thirds of multiple diseases may be tested for under the same code, now claims used nonspecific codes. In addition, tracking the use reflect only the laboratory methods used for running the test.8 ofmanynewersinglegenetests,genepanels,andexome/ genome sequencing was not possible. A minority of patients Unfortunately, it was not possible to determine diagnostic in- analyzed had a neurology-related ICD-9 CCS associated tent through analysis of tier 2 codes, which represented 3,637

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 (8.2%) of tests in our study. This finding is consistent with previously reported use of tier 2 codes in a population of Appendix Authors 9 Medicare patients, suggesting that the root cause of the am- Name Location Role Contribution biguity lies in the codes themselves and not the databases Samuel J. University of Author Designed and directed the housing the resulting claims. Of interest, although tier 1 codes Mackenzie, Michigan, Ann research project; analyzed were initially intended to represent tests completed at high MD, PhD Arbor, MI the data; participated in the data interpretation; created volume, claims submitted for some tier 1 conditions such as figure; drafted, reviewed, Fragile X (367; 0.8%), Duchenne muscular dystrophy (12; and approved the final <0.1%), and Charcot-Marie-Tooth (PMP22; 61; 0.1%) were manuscript observed with relatively low frequency. Tier 1 codes pertaining Chun Chieh University of Author Collected and analyzed Lin, PhD, Michigan, Ann data; participated in the to a possible underlying coagulopathy diagnosis predisposing MBA Arbor, MI data interpretation to cerebrovascular disease were used more commonly. reviewed and approved the final manuscript We hypothesize that several tests with specific corresponding codes may have been coded under nonspecific designations Peter K. University of Author Participated in the data Todd, MD, Michigan, Ann interpretation; reviewed instead. As an example, whole exome sequencing, a technique PhD Arbor, MI and approved the final that has been demonstrated to have a high diagnostic yield in manuscript a number of clinical contexts within neurology, including James F. University of Author Designed and directed the ataxia, autism, and intellectual disability,10 was identifiable Burke, MD, Michigan, Ann research project; fi MS Arbor, MI participated in the data through its speci c codes (81215 and 81416) in only 3 cases. interpretation; reviewed In addition, a significant amount of other (nongenetic) di- and approved the final agnostic testing was likely captured through nonspecific codes manuscript such as 89240 (unlisted miscellaneous pathology test). Brian C. University of Author Designed and directed the Callaghan, Michigan, Ann research project; MD, MS Arbor, MI participated in the data Accurate population-based utilization data may help identify interpretation; reviewed disparities in genetic testing among certain patient populations, and approved the final manuscript determine total and out-of-pocket costs associated with genetic testing, elucidate how often certain molecular targets may be being redundantly probed by way of serial testing, and identify ff References ways to optimize time to diagnosis in a cost-e ective manner. 1. Klein CJ, Foroud T. Neurology individualized medicine: when to use next-generation Our study suggests, however, that traditional claims-based sequencing panels. Mayo Clin Proc 2017;92:292–305. 2. Riley JD, Procop GW, Kottke-Marchat K, Wyllie R, Lacbawan FL. Improving mo- methods have limited application to genetic testing at present, lecular genetic test utilization through order restriction, test review, and guidance. given the nonspecificity of unlisted and tier 2 molecular pa- J Mol Diagn 2015;17:225–229. 3. Biesecker LG, Green RC. Diagnostic clinical genome and exome sequencing. N Engl J thology codes. Given the speed with which new genetic tests Med 2014:2418–2425. come into and out of the market, this is a problem that warrants 4. Salm M, Abbate K, Appelbaum P, et al. Use of genetic tests among neurologists and psychiatrists: knowledge, attitudes, behaviors, and needs for training. J Genet Couns careful consideration going forward. Potential solutions will 2014;23:156–163. likely require input from multiple stakeholders, including 5. National Institutes of Health Genetic Testing Registry [online]. Available at: ncbi. health systems, payers, and commercial laboratories. nlm.nih.gov/gtr/. Accessed July 25, 2019. 6. Hsiao SJ, Mansukhani MM, Carter MC, Sireci AN. The history and impact of molecular coding changes on coverage and reimbursement of molecular di- Study funding agnostic tests: transition from stacking codes to the current molecular code set including genomic sequencing procedures. J Mol Diagn 2018;20: No targeted funding reported. 177–183. 7. Klein RD. Reimbursement in molecular pathology: bringing genomic medicine to patients. Clin Chem 2015;61:136–138. Dislosure 8. CPT molecular pathology tier 2 codes [online]. Available at: ama-assn.org/sites/ama- Disclosures available: Neurology.org/NG. assn.org/files/corp/media-browser/public/physicians/cpt/cpt-molecular-pathol- ogy-tier-2-codes.pdf. Accessed February 22, 2019. 9. Lynch JA, Berse B, Dotson WD, Khoury J, Coomer N, Kautter J. Utilization of genetic Publication history tests: analysis of gene-specific billing in Medicare claims data. Genet Med 2017;19: – fi 890 899. Received by Neurology: Genetics September 23, 2019. Accepted in nal 10. Fogel BL, Satya-Murti S, Cohen BH. Clinical exome sequencing in neurologic disease. form January 9, 2020. Neurol Clin Pract 2016;6:164–176.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG ARTICLE OPEN ACCESS Association of a structural variant within the SQSTM1 gene with amyotrophic lateral sclerosis

Julia Pytte, BSc (Hons), Ryan S. Anderton, PhD, Loren L. Flynn, PhD, Frances Theunissen, MBiomedSc, Correspondence Leanne Jiang, BSc (Hons), Ianthe Pitout, PhD, Ian James, PhD, Frank L. Mastaglia, MD, Ann M. Saunders, PhD, Prof. Akkari [email protected] Richard Bedlack, MD, PhD, Teepu Siddique, MD, Nailah Siddique, RN, Msn, and P. Anthony Akkari, PhD

Neurol Genet 2020;6:e406. doi:10.1212/NXG.0000000000000406 Abstract Objective As structural variations may underpin susceptibility to complex neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS), the objective of this study was to investigate a structural variant (SV) within sequestosome 1 (SQSTM1).

Methods A candidate insertion/deletion variant within intron 5 of the SQSTM1 gene was identified using a previously established SV evaluation algorithm and chosen according to its subsequent theoretical effect on gene expression. The variant was systematically assessed through PCR, polyacrylamide gel fractionation, Sanger sequencing, and reverse transcriptase PCR.

Results A reliable and robust assay confirmed the polymorphic nature of this variant and that the variant may influence SQSTM1 transcript levels. In a North American cohort of patients with familial ALS (fALS) and sporadic ALS (sALS) (n = 403) and age-matched healthy controls (n = 562), we subsequently showed that the SQSTM1 variant is associated with fALS (p = 0.0036), particularly in familial superoxide dismutase 1 mutation positive patients (p = 0.0005), but not with patients with sALS (p = 0.97).

Conclusions This disease association highlights the importance and implications of further investigation into SVs that may provide new targets for cohort stratification and therapeutic development.

From the University of Western Australia (J.P., R.S.A., L.L.F., F.T., L.J., F.L.M., P.A.A.), Centre for Neuromuscular and Neurological Disorders, Crawley; Perron Institute for Neurological and Translational Science (J.P., R.S.A., L.L.F., F.T., L.J., I.P., F.L.M., P.A.A.), Nedlands; University of Notre Dame Australia (R.S.A.), School of Health Sciences; University of Notre Dame Australia (R.S.A.), Institute for Health Research, Fremantle; Murdoch University (L.L.F., I.P., P.A.A.), Centre for Molecular Medicine and Innovative Therapeutics; Murdoch University, Institute for Immunology and Infectious Diseases (I.J.), Western Australia, Australia; Department of Neurology (R.B.), Duke University School of Medicine, Durham, NC; Zinfandel Pharmaceuticals (A.M.S.), Inc.; Duke University (R.B.), ALS Clinic, Durham, NC; and Departments of Neurology, Pathology and Cell and Molecular Biology (T.S., N.S.), Northwestern University Feinberg School of Medicine, the Les Turner ALS Center and the Northwestern University Interdepartmental Neuroscience Program, Chicago, IL.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary ALS = amyotrophic lateral sclerosis; CI = confidence interval; fALS = familial ALS; GWAS = genome-wide association study; I/ D = insertion/deletion; mRNA = messenger RNA; ONS = olfactory neurosphere derived; OR = odds ratio; sALS = sporadic ALS; SOD1 = superoxide dismutase 1; SQSTM1 = sequestosome 1; RT-PCR = reverse transcriptase PCR; SV = structural variant.

A growing body of literature indicates that structural variants Insurance Portability and Accountability Act Standards of (SVs) are important in complex diseases because their role in Confidentiality and Disclosure and approved by the North- gene expression and messenger RNA (mRNA) regulation is western University institutional review board (STU0012722/ – now emerging.1 4 Given the potential implication of SVs in CR5_STU00012722) and the Duke University institutional – neurodegenerative disorders,5 7 an investigation of SVs within review board (PRO00040665/323682). All patients were di- associated genes may provide insight toward the pathogenic agnosed by board-certified neurologists and met the revised El mechanisms involved and enable the identification of new Escorial World Federation of Neurology criteria for diagnosis therapeutic targets. with ALS.19

Mutations in sequestosome 1 (SQSTM1), and aggregation of SV identification the SQSTM1-encoded p62 protein, have been identified in An SV evaluation algorithm was used to locate the variant patients with amyotrophic lateral sclerosis (ALS) and other within the gene, SQSTM1.18 Detailed methods are described – neurodegenerative diseases.8 13 P62 is a multifunctional protein in the supplementary material (see e-methods, links.lww. that binds ubiquitin and is involved in autophagy, proteasomal com/NXG/A229). degradation of ubiquitinated proteins, mitophagy, and cellular – signaling.14 16 Variants within SQSTM1 or surrounding PCR amplification and Sanger sequencing SQSTM1 may contribute to the diverse presentation observed Detailed methods are described in the supplementary mate- between the patients with ALS. How mutations in SQSTM1 rial. Sequences of all primers used for this study are outlined in influence ALS is not fully understood; however, it is believed table e-1 (links.lww.com/NXG/A229). that structural changes to the protein may affect adapter func- tion of SQSTM1 to the LC3 protein in the nascent autopha- Olfactory neurosphere-derived cells gosome and impair autophagy of proteins that are not recyclable Culturing information can be found in the supplementary data. by the proteasome.17 Other cellular systems may also be affected such as ubiquitin binding and regulation of cellular processes RT-PCR and densitometry including DNA repair, endocytosis, and signal transduction.8,14 Detailed methods are described in the supplementary mate- rial. Sequences of all primers used for this study are detailed in In the current study, a potential influential insertion/deletion table e-1 (links.lww.com/NXG/A229). (I/D) within intron 5 of SQSTM1 was identified using an in silico short SV evaluation algorithm.18 Reverse transcriptase ALS and healthy control participants PCR (RT-PCR) analysis revealed a link between the I/D and A cohort of 196 fALS, 207 sALS, and 562 healthy control altered transcript levels. An exploratory association study in participants were recruited into the Neurologic Diseases a sporadic and familial North American ALS cohort established Registry, Northwestern University, Chicago, USA, and Duke University, North Carolina, USA. The 196 fALS cases an association with familial ALS (fALS), particularly superoxide ff dismutase 1 (SOD1) mutation-positive patients but not spo- from Northwestern University belong to 74 di erent fam- radic ALS (sALS). We further examined whether this SV was ilies with possible within-family genetic correlations. The associated with age of disease onset and duration in SOD1 mutation data for all fALS patients are detailed in table e-2 mutation-positive patients and found no statistically significant (links.lww.com/NXG/A229). association. Statistics Data are reported as mean ± SD where appropriate. Statistical Methods differences in genotype proportions for independent cases were assessed using Pearson χ2 test with Yates correction. Proportions Standard protocol approvals, registrations, and ages of onset involving familial groups were analyzed using and patient consents the mixed effects regression models to account for possible This study was approved by the ethical standards of the relevant within-family correlations. Durations were analyzed using Cox institutional review board, the Human Research Ethics Com- proportional hazards models with clustering to account for the mittee of the University of Western Australia (RA/4/20/ correlations. Analyses were carried out in IBM SPSS Statistics 5308). Participants were enrolled after informed consent was version 25.0 (IBM Co., Armonk, NY) and R version 3.4.3 obtained. Clinical data were collected according to the Health (R Foundation for Statistical Computing, Vienna, Austria).

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Data availability Study data for the primary analyses presented in this report are Figure 1 Characterization of the SQSTM1 variant available on reasonable request from the corresponding author.

Results Identification of a polymorphic variant within the SQSTM1 gene An SV evaluation algorithm was used to identify the genetic variant located within SQSTM1.Theidentified cytosine adenine adenine adenine (CAAA) I/D within intron 5 of SQSTM1 was evaluated on the NCBI database; although the region had been previously reported (NC_0000005.0), no disease associations were established. The recorded NCBI data demonstrated un- certainty to its polymorphic nature; currently, the region is recorded as a series of insertion/deletions of varying length, with no allelic frequency data. Conventional polyacrylamide gel electrophoresis revealed a biallelic variant, consisting of a het- erozygous I/D genotype in lanes 2, 3, and 7, a homozygous D/ D genotype in lanes 4 and 8, and a homozygous I/I genotype in lanes5,6,and9(figure 1A). Sanger sequencing confirmed that the I allele contained a CAAA insertion when compared with the D allele (figure 1B). To determine the effect of the CAAA SV on SQSTM1 expression, SQSTM1 was assessed in a panel (n = 3) of olfactory neurosphere-derived (ONS) cells obtained from the healthy controls. Semiquantitative RT-PCR revealed a stepwise increase in the level of SQSTM1 transcript, with the I/I genotype yielding the lowest levels and the I/D and D/D genotypes producing a 1.77-fold and 2.47-fold relative increase in SQSTM1 transcript, respectively (figure 1C). (A) PCR and native polyacrylamide gel electrophoresis across a random selec- tion of control DNA samples and compared against a 100bp ladder. (B) Sanger The SQSTM1 variant is associated with fALS sequencing of the I allele and D allele. (C) SQSTM1 transcript levels of SQSTM1 exon 4–7 analyzed using RT-PCR on RNA from a panel of control ONS cells. Following the identification of the CAAA I/D variant, we Relative densitometry was calculated with SQSTM1 transcript signal standard- proceeded to determine the variant frequency in a cohort of ized to each respective GAPDH signal. GAPDH = glyceraldehyde 3-phosphate dehydrogenase; I/D = insertion/deletion; ONS = olfactory neurosphere derived 196 fALS patients, 207 sALS patients, and 562 healthy age- cells; RT-PCR = Reverse transcriptase PCR; SQSTM1 = sequestosome 1. matched controls. The characteristics of the study participants are summarized in table 1, including sex, age, disease duration, and family history of each cohort. Of the ALS cohort, patients SQSTM1 association was stronger when examining the subset with familial mutations (48.6%) were further grouped by af- of fALS patients carrying mutations in SOD1 (χ2 =16.754, fected gene, specifically C9orf72 (2.5% of patients with ALS), df = 2, p = 0.0002, OR = 1.869, 95% CI: 1.281, 2.726 assuming SOD1 (41.4% of patients with ALS), or TDP-43 (4.7% of independent cases). When accounting for family structures, the patients with ALS). SQSTM1 variant was strongly associated with patients carrying a SOD1 mutation both genotypically and allelically (p =0.001 There was no difference in the frequency of the SQSTM1 and p=0.0005, respectively) (table 2). When analyzed with variant observed between healthy controls and sALS cases (χ2 SOD1-A5V mutation-positive patients excluded, the strong = 0.032, df = 2, p = 0.984, odds ratio [OR] = 0.974, 95% SOD1 association remains (χ2 = 6.801, p < 0.009, OR = 1.830, confidence interval [CI]: 0.660, 1.436). However, there was 95% CI: 1.157, 2.896). adifference in SQSTM1 variant frequency observed between healthy controls and fALS patients (χ2 =12.791,df=2,p = The SQSTM1 variant is not associated with age at onset of 0.002, OR = 1.79, 95% CI: 1.250, 2.562 assuming in- disease or survival in patients with SOD1. dependence of cases), with the homozygous (I/I) genotype over-represented in the fALS cohort (33.2%) when compared To determine if the SQSTM1 variant is associated with SOD1 with the controls (21.7%). When analyzed using mixed effects mutation-positive ALS patient outcomes, age at onset and regression models to account for possible within-family cor- survival were analyzed. There was no association observed relations, the (I/I) association remained both genotypical and between the age at onset in patients with the I/I (Mean Rank = allelical (p = 0.013 and p=0.0036, respectively, table 2). This 87.25), I/D (Mean Rank = 76.40), or D/D (Mean Rank =

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Table 1 Characteristics of patients and healthy control Table 2 Association of SQSTM1 variant and ALS disease participants status

Mean (SD) or n sALS Control n (%) sALS n (%) p Value

Patients Controls SQSTM1 CAAA variant Variable (n = 403) (n = 562) I/I 122 (21.7) 44 (21.3) Sex I/D 292 (52.0) 109 (52.7) 0.98 Male 208 296 D/D 148 (26.3) 54 (26.1) Female 195 266 I allele 536 197 Age (y) 52.38 (12.99) 49.37 (12.56) D allele 588 217 0.97 Disease duration (mo) 43.21 (46.98) — Familial ALS Control n (%) fALS n (%) p Valuea Family history — SQSTM1 CAAA variant Sporadic 207 I/I 122 (21.7) 65 (33.2) 0.0013 Familial 196 — I/D 292 (52.0) 97 (49.5) 0.50

D/D 148 (26.3) 34 (17.3) 0.08 fi 84.43) genotypes (p = 0.2, mixed model regression, gure 2A). I allele 536 227 Survival curves were generated to compare durations of each genotypic group (figure 2B). No association was observed D allele 588 165 0.0036 between patients carrying each genotype and their survival after SOD1 mutation a taking into account of familial correlations via a clustered Cox positive Control n (%) SOD1 n (%) p Value regression (p = 0.55). SQSTM1 CAAA variant

I/I 122 (21.7) 57 (34.1) 0.001 Discussion I/D 292 (52.0) 87 (52.1) 0.99 A growing body of literature has highlighted SVs, their abun- D/D 147 (26.3) 23 (13.8) 0.014 dance throughout the human genome, and their potential role I allele 536 201 in the pathogenesis of ALS and other neurodegenerative – D allele 588 133 0.0005 diseases.20 24 SVs are responsible for greater diversity at the nucleotide level between 2 human genomes than any other Abbreviations: ALS = amyotrophic lateral sclerosis; fALS = familial ALS; I/D = insertion/deletion; sALS = sporadic ALS; SQSTM1 = sequestosome 1. form of genetic variations and are three-fold more likely to a Each row compared using random effects to account for familial associate with genome-wide association studies (GWASs) correlation. signals than single nucleotide polymorphisms (SNPs).20 SVs that remain cryptic to current sequencing algorithms are likely to account for disease-causing variation in unsolved Mendelian disorders and missing heritability in complex inclusions are found in virtually all forms of ALS and ALS- disorders.20,23 SVs may affect gene expression and therefore frontotemporal dementia.8 We now provide a third link of may play an important but understudied role in disease SQSTM1 variants to the fALS. SQSTM1-encoded p62 protein is susceptibility.20,25 Recent discoveries of SVs as informative akeyscaffolder involved in cellular signaling and protein deg- disease risk markers for rare genetic disorders provide radation through the autophagosome-lysosome system.17 compelling evidence for ongoing investigation into the as- SQSTM1/p62 mutations may confer a toxic gain of function – sociation between SVs and rare genetic diseases.26 30 The through protein interactions, leading to dysregulation of cell novel bioinformatics SV evaluation algorithm tool prioritizes signaling pathways, protein misfolding, and aggregation.8 Evi- potential functional/causal SVs within candidate regions dence for this in the literature suggests parallels between p62 identified using GWAS.18 In ALS and all rare genetic dis- and other proteins associated with neurodegeneration, in- eases, these highly polymorphic markers are often over- cluding SOD1-linked ALS, C9orf72, ubiquilin 2, TDP-43, FUS, looked, largely because of the limitations in the current gene optineurin, beta-amyloid, α-synuclein, and tau.8,31 Owing to its sequencing platforms such as next-generation sequencing role in protein degradation, overexpression of p62 has been and GWAS, which are primarily designed to detect SNPs. showntobeprotectiveinsomeneurodegenerativeanimal models, but overexpression in a SOD1 ALS model was found to – Two direct lines of evidence link SQSTM1 to ALS: the first accelerate disease onset.32 34 Taken together, these studies through etiology on account of mutations associated with ALS suggest that a fine balance in p62 levels is required for optimal and second through pathology, where p62 immune-reactive signaling and protein clearance. Consequently, small changes in

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG aggregation, have been reported not only in ALS but also in Figure 2 Association of the SQSTM1 variant with age at other complex neurodegenerative diseases, including Par- onset of disease and survival kinson disease, Paget disease of bone, Alzheimer disease and – frontotemporal dementia.8 13 A semiquantitative assess- ment of the mRNA expression revealed a difference in SQSTM1 transcript levels in the ONS cells carrying each genotype. Neural tissue derived from olfactory mucosa, such as primary ONS cells, provide informative cellular models for neurodegenerative diseases. Primary ONS cells are neural progenitor cells and more accurately reflect motor neuron cellular function than fibroblasts and PBMCs, and unlike postmortem brain tissue, they can be isolated from living patients.37,38 As such, the altered SQSTM1 transcript levels observed in the ONS cells with the I/I genotype may be reflective of altered SQSTM1 expression in motor neu- rons. Although this result requires validation in a larger sample size, it does suggest some level of regulation by the SV on SQSTM1 transcript expression. This may be due to altered transcription efficiency, splicing, or transcript sta- bility, potentially translating to altered protein expression that may contribute to disease pathologies.

An evaluation of the SQSTM1 variant within a cohort of 403 patients with ALS and 562 healthy age-matched controls revealed an association with fALS disease risk, but not age of disease onset or disease severity. Replications in additional fALS cohorts are necessary to truly elucidate the nature of this variant. Causality has been reported between ALS and SQSTM1 variants rs796051870, rs776749939, rs796052214, and as such, the ongoing investigation between ALS and SQSTM1 gene variants is necessary.8,39,40 P62 has previously been reported as overexpressed and accumulated in inclusions of sporadic inclusion body myositis, reinforcing its importance in neurodegenerative diseases.12 We anticipate that as addi- tional SVs are identified, these will further stratify other relevant disease phenotypes such as age at onset, duration, and disease progression.

Acknowledgment (A) The median and distribution of age at onset (years) of SOD1 mutation- positive patients with ALS (n = 167) grouped by each SQSTM1 variant geno- The authors are deeply indebted to the late Allen D. Roses, type. (B) Kaplan-Meier survival curves of SOD1 mutation-positive patients, PhD, who provided mentorship to several of the authors and comparing the SQSTM1 genotypes assuming independent measurements. A robust log-rank test accounting for familial correlation was performed to who originally conceptualized and initiated this research. The assess any association between the groups. Survival was measured in authors would also like to thank the contribution of Prof. Alan months from ALS diagnosis until death. ALS = amyotrophic lateral sclerosis; SOD1 = superoxide dismutase 1; SQSTM1 = sequestosome 1. Mackay Sim and Prof. George Mellick (Griffith University, Institute for Drug Discovery) for the generous gift of the ONS cell cultures. The study was in part funded by NIH grants to TS the level of p62 expression, whichmaybeproducedbyvariants (NS050641, NS046535), the Les Turner ALS Foundation/ such as the SQSTM1 intron 5 I/D, could tip the balance of p62 Herbert C. Wenske Foundation Professorship and the Les expression, contributing to the disease. Previously identified Turner ALS Research and Patient Care Center. mutations within SQSTM1 have been associated with ALS.35,36 As such, it has been suggested that these mutations may have Study funding a direct role in ALS pathogenesis, presenting as an important The study was funded by research support from the Perron target for future therapy.8 Institute for Neurologic and Translational Science and the Les Turner Foundation for ALS. The funders have no role in In this study, we hypothesized that structural variations in the design of the study and collection, analysis, decision SQSTM1 may uncover novel susceptibility factors that un- to publish, interpretation of data, or preparation of the derpin this disease. SQSTM1 mutations, and p62 manuscript.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Dislosure Disclosures available: Neurology.org/NG. Appendix (continued)

Name Location Contribution Publication history Received by Neurology: Genetics July 19, 2019. Accepted in final form Richard Duke University, Durham, Study concept and Bedlack, MD, NC design, acquisition of January 23, 2020. PhD the data, and critical revision of the manuscript for important intellectual content

Appendix Authors Teepu Northwestern University, Study concept and Siddique, MD, Evanston, IL design, acquisition of Name Location Contribution DSc (hc), the data, and critical FAAN revision of the Julia Pytte, The Perron Institute for Acquisition of the data, manuscript for BSc (Hons) Neurological and analysis and important intellectual Translational Science, interpretation, content Nedlands, Australia statistical analysis, and critical revision of the Nailah Northwestern University, Study concept and manuscript for Siddique, RN, Evanston, IL design and critical important intellectual Msn revision of the content manuscript for important intellectual Ryan S. The Perron Institute for Study concept and content Anderton, Neurological and design, statistical PhD Translational Science, analysis, study P. Anthony The Perron Institute for Study concept and Nedlands, Australia supervision, and critical Akkari, PhD Neurological and design, acquisition of revision of the Translational Science, data, analysis and manuscript for Crawley, Australia interpretation, and important intellectual critical revision of the content manuscript for important intellectual Loren L. Murdoch University, Study concept and content and study Flynn, PhD Murdoch, Australia design, analysis and supervision interpretation, study supervision, and critical revision of the manuscript for important intellectual References content 1. Chakravarti A, Kapoor A. Mendelian puzzles. Science 2012;335:930–931. 2. Feuk L, Carson AR, Scherer SW. Structural variation in the human genome. Nat Rev Frances The Perron Institute for Analysis and Genet 2006;7:85–97. Theunissen, Neurological and interpretation and 3. Stankiewicz P, Lupski JR. Structural variation in the human genome and its role in MBiomedSc Translational Science, critical revision of the disease. Annu Rev Med 2010;61:437–455. Nedlands, Australia manuscript for 4. Lupski JR. Structural variation mutagenesis of the human genome: impact on disease important intellectual and evolution. Environ Mol Mutagen 2015;56:419–436. content 5. Lyall DM, Royle NA, Harris SE, et al. Alzheimer’s disease susceptibility genes APOE and TOMM40, and hippocampal volumes in the Lothian birth cohort 1936. PLoS Leanne Jiang, The Perron Institute for Analysis and One 2013;8:e80513. BSc (Hons) Neurological and interpretation and 6. Lutz MW, Saul R, Linnertz C, et al. A cytosine-thymine (CT)-rich haplotype in intron Translational Science, critical revision of the 4 of SNCA confers risk for Lewy body pathology in Alzheimer’s disease and affects Nedlands, Australia manuscript for SNCA expression. Alzheimers Dement 2015;11:1133–1143. important intellectual 7. Linnertz C, Anderson L, Gottschalk W, et al. The cis-regulatory effect of an Alz- content heimer’s disease-associated poly-T locus on expression of TOMM40 and apolipo- protein E genes. Alzheimers Dement 2014;10:541–551. Ianthe Pitout, The Perron Institute for Study concept and 8. Fecto F, Yan J, Vemula SP, et al. SQSTM1 mutations in familial and sporadic PhD Neurological and design and acquisition amyotrophic lateral sclerosis. Arch Neurol 2011;68:1440–1446. Translational Science, of the data 9. Rea SL, Majcher V, Searle MS, Layfield R. SQSTM1 mutations—bridging paget Nedlands, Australia disease of bone and ALS/FTLD. Exp Cell Res 2014;325:27–37. 10. Zatloukal K, Stumptner C, Fuchsbichler A, et al. p62 Is a common component of Ian James, Murdoch University, Statistical analysis and cytoplasmic inclusions in protein aggregation diseases. Am J Pathol 2002;160: PhD Murdoch, Australia critical revision of the 255–263. manuscript for 11. Lowe J, Blanchard A, Morrell K, et al. Ubiquitin is a common factor in intermediate important intellectual filament inclusion bodies of diverse type in man, including those of Parkinson’s content disease, Pick’s disease, and Alzheimer’s disease, as well as Rosenthal fibres in cerebellar astrocytomas, cytoplasmic bodies in muscle, and mallory bodies in alcoholic liver Frank The Perron Institute for Critical revision of the disease. J Pathol 1988;155:9–15. Mastaglia, Neurological and manuscript for 12. Nogalska A, Chiara AE, Ae T, et al. p62/SQSTM1 is overexpressed and prominently MD Translational Science, important intellectual accumulated in inclusions of sporadic inclusion-body myositis muscle fibers, and can Nedlands, Australia content help differentiating it from polymyositis and dermatomyositis. Acta Neuropathol 2009;118:407–413. Ann M. Duke University, Durham, Study concept and 13. Kwok CT, Morris A, de Belleroche JS. Sequestosome-1 (SQSTM1) sequence variants Saunders, NC design and critical in ALS cases in the UK: prevalence and coexistence of SQSTM1 mutations in ALS PhD revision of the kindred with PDB. Eur J Hum Genet 2014;22:492–496. manuscript for 14. Kirkin V, McEwan DG, Novak I, Dikic I. A Role for ubiquitin in selective autophagy. important intellectual Mol Cell 2009;34:259–269. content 15. Welchman RL, Gordon C, Mayer RJ. Ubiquitin and ubiquitin-like proteins as mul- tifunctional signals. Nat Rev Mol Cell Biol 2005;6:599–609.

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG 16. Rogov V, D¨otsch V, Johansen T, Kirkin V. Interactions between autophagy receptors 28. Renton AE, Majounie E, Waite A, et al. A hexanucleotide repeat expansion in C9ORF72 and ubiquitin-like proteins form the molecular basis for selective autophagy. Mol Cell is the cause of chromosome 9p21-linked ALS-FTD. Neuron 2011;72:257–268. 2014;53:167–178. 29. Sproviero W, Shatunov A, Stahl D, et al. ATXN2 trinucleotide repeat length correlates 17. Pankiv S, Clausen TH, Lamark T, et al. p62/SQSTM1 binds directly to Atg8/LC3 to with risk of ALS. Neurobiol Aging 2017;51:178.e-1–178. facilitate degradation of ubiquitinated protein aggregates by autophagy. J Biol Chem 30. Roses AD, Lutz MW, Amrine-Madsen H, et al. A TOMM40 variable-length poly- 2007;282:24131–24145. morphism predicts the age of late-onset Alzheimer’s disease. Pharmacogenomics J 18. Saul R, Lutz MW, Burns DK, Roses AD, Chiba-Falek O. The SSV evaluation system: 2010;10:375–384. a tool to prioritize short structural variants for studies of possible regulatory and causal 31. Deng HX, Chen W, Hong ST, et al. Mutations in UBQLN2 cause dominant X-linked variants. Hum Mutat 2016;37:877–883. juvenile and adult-onset ALS and ALS/dementia. Nature 2011;477:211–215. 19. Brooks BR, Miller RG, Swash M, Munsat TL, World Federation of Neurology Re- 32. Doi H, Adachi H, Katsuno M, et al. p62/SQSTM1 differentially removes the toxic search Group on Motor Neuron Diseases. El Escorial revisited: revised criteria for the mutant androgen receptor via autophagy and inclusion formation in a spinal and diagnosis of amyotrophic lateral sclerosis. Amyotroph Lateral Scler Other Mot bulbar muscular atrophy mouse model. J Neurosci 2013;33:7710–7727. Neuron Disord 2000;1:293–299. 33. Hadano S, Mitsui S, Pan L, et al. Functional links between SQSTM1 and ALS2 in the 20. Chaisson MJP, Sanders AD, Zhao X, et al. Multi-platform discovery of haplotype- pathogenesis of ALS: cumulative impact on the protection against mutant SOD1- resolved structural variation in human genomes. Nat Commun 2019;10:1784. mediated motor dysfunction in mice. Hum Mol Genet 2016;25:3321–3340. 21. Sudmant PH, Rausch T, Gardner EJ, et al. An integrated map of structural variation in 34. Mitsui S, Otomo A, Nozaki M, et al. Systemic overexpression of SQSTM1/p62 2,504 human genomes. Nature 2015;526:75–81. accelerates disease onset in a SOD1 H46R-expressing ALS mouse model. Mol Brain 22. Roses AD. Short sequence variations (SSVs) may explain multiplicity of SOD1-SNP 2018;11:1–16. mutations in Amyotrophic Lateral Sclerosis. Presented at the 66th Annual Meeting 35. Gal J, Strom A-L, Kwinter DM, et al. Sequestosome 1/p62 links familial ALS mutant American Society of Human Genetics; October 19, 2016; Vancouver. SOD1 to LC3 via a ubiquitin-independent mechanism. J Neurochem 2009;111: 23. Roses AD, Akkari PA, Chiba-Falek O, et al. Structural variants can be more in- 1062–1073. formative for disease diagnostics, prognostics and translation than current SNP 36. Zhou L, Wang H, Ren H, et al. Bcl-2-dependent upregulation of autophagy by mapping and exon sequencing. Expert Opin Drug Metab Toxicol 2016;12:135–147. sequestosome 1/p62 in vitro. Acta Pharmacol Sin 2013;34:651–656. 24. Frazer KA, Murray SS, Schork NJ, Topol EJ. Human genetic variation and its con- 37. F´eron F, Perry C, Girard SD, Mackay-Sim A. Isolation of adult stem cells from the tribution to complex traits. Nat Rev Genet 2009;10:241–251. human olfactory mucosa. In: Reynolds B, Deleyrolle L, editors. Methods in Molecular 25. Li YC, Korol AB, Fahima T, Nevo E. Microsatellites within genes: structure, function, Biology. New Jersey: Humana Press; 2013:107–113. and evolution. Mol Biol Evol 2004;21:991–1007. 38. Matigian N, Abrahamsen G, Sutharsan R, et al. Disease-specific, neurosphere-derived 26. De Jesus-Hernandez M, Mackenzie IR, Boeve BF, et al. Expanded GGGGCC hex- cells as models for brain disorders. Dis Model Mech 2010;3:785–798. anucleotide repeat in noncoding region of C9ORF72 causes chromosome 9p-Linked 39. Teyssou E, Takeda T, Lebon V, et al. Mutations in SQSTM1 encoding p62 in FTD and ALS. Neuron 2011;72:245–256. amyotrophic lateral sclerosis: genetics and neuropathology. Acta Neuropathol 2013; 27. Gijselinck I, Van Langenhove T, van der Zee J, et al. A C9orf72 promoter repeat 125:511–522. expansion in a Flanders-Belgian cohort with disorders of the frontotemporal lobar 40. Le Ber I, Camuzat A, Guerreiro R, et al. SQSTM1 mutations in French patients with degeneration-amyotrophic lateral sclerosis spectrum: a gene identification study. frontotemporal dementia or frontotemporal dementia with amyotrophic lateral Lancet Neurol 2012;11:54–65. sclerosis. JAMA Neurol 2013;70:1403–1410.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 ARTICLE OPEN ACCESS Novel EGR2 variant that associates with Charcot-Marie-Tooth disease when combined with lipopolysaccharide-induced TNF-α factor T49M polymorphism

Maria Empar Blanco-Canto,´ MD,* Nikiben Patel, MSc,* Sergio Velasco-Aviles, MSc, Angeles Casillas-Bajo, BSc, Correspondence Juan Salas-Felipe, MD, Alexandre Garc´ıa-Escriv´a, MD, Carmen D´ıaz-Mar´ın, MD, PhD, and Dr. Cabedo [email protected] Hugo Cabedo, MD, PhD or Dr. D´ıaz-Mar´ın [email protected] Neurol Genet 2020;6:e407. doi:10.1212/NXG.0000000000000407 Abstract Objective To identify novel genetic mechanisms causing Charcot-Marie-Tooth (CMT) disease.

Methods We performed a next-generation sequencing study of 34 genes associated with CMT in a pa- tient with peripheral neuropathy.

Results We found a non–previously described mutation in EGR2 (p.P397H). P397H mutation is located within the loop that connects zinc fingers 2 and 3, a pivotal domain for the activity of this transcription factor. Using promoter activity luciferase assays, we found that this mutation promotes decreased transcriptional activity of EGR2. In this patient, we also found a previously described nonpathogenic polymorphism in lipopolysaccharide-induced TNF-α factor (LITAF) (p.T49M). We show that the p.T49M mutation decreases the steady-state levels of the LITAF protein in Schwann cells. Loss of function of LITAF has been shown to produce deregulation in the NRG1-erbB signaling, a pivotal pathway for EGR2 expression by Schwann cells. Surpris- ingly, our segregation study demonstrates that p.P397H mutation in EGR2 is not sufficient to produce CMT disease. Most notably, only those patients expressing simultaneously the LITAF T49M polymorphism develop peripheral neuropathy.

Conclusions Our data support that the LITAF loss-of-function interferes with the expression of the transcriptional-deficient EGR2 P397H mutant hampering Schwann cell differentiation and suggest that in vivo both genes act in tandem to allow the proper development of myelin.

*These authors contributed equally to this work.

From the ISABIAL (FISABIO) (M.E.B.-C., N.P., S.V.-A., A.C.-B., C.D.-M., H.C.), Hospital General Universitario de Alicante; Instituto de Neurociencias de Alicante UMH-CSIC (N.P., S.V.-A., A.C.-B., H.C.), San Juan de Alicante, Spain; Hospital Marina Salud (J.S.-F.), Denia; and Hospital IMED Levante (A.G.-E.), Benidorm, Spain.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by ISABIAL. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glosary ChIP = chromatin immunoprecipitation; CMT = Charcot-Marie-Tooth; CMTNS = CMT neuropathy score; LITAF = lipopolysaccharide-induced TNF-α factor.

Many genes responsible for Charcot-Marie-Tooth (CMT) dis- Plasmids ease have been identified to date, and it has been estimated that at pGFP-hLITAF and pEGFP-hLITAF T49M were provided by least other 30 CMT-causing genes could be still discovered in the Dr. Philip Woodman and Dr. Lydia Wunderley (University of – near future.1 3 EGR2 (also known as KROX20) is a transcription Manchester, UK). Where necessary, the DNA encoding for factor that activates the myelination program in Schwann cells.4,5 green flourescent protein (GFP) was removed with restriction Different alterations in the zinc finger region of this transcription enzymes using standard molecular biology cloning methods to factor lead to demyelination causing the autosomal dominant generate phLITAF and phLITAF T49M. A PMP22 intronic CMT1 type D.6 The lipopolysaccharide-induced TNF-α factor enhancer cloned in pGL4 (phPMP22enh-Luciferase) and (LITAF) is an 18-kDa protein involved in endosomal recycling pcDNA3.1 hEGR2 construct were provided by Dr. John Sva- and protein degradation. Eight missense mutations in the LITAF ren (University of Wisconsin). pcDNA3.1-EGR2 P397H was gene have been associated with autosomal dominant CMT1 type generated with the Quikchange Site-Directed Mutagenesis Kit C.7 Although some of these mutations cause mislocation of (Stratagene, La Jolla, CA). The correct sequences of the con- LITAF from the lysosomal membrane to the cytosol and others struct were confirmed by sequencing. induce the formation of aggresomes,8,9 the exact pathogenic mechanism inducing demyelination remains unknown. Here, we Cell cultures describe a family with a novel mutation in EGR2 (EGR2 P397H). Schwann cells were cultured from sciatic nerves of neonatal Although the mutation reduces the capacity of this transcription rats as described previously11 with minor modifications (see factor to induce myelin gene expression in vitro, in human het- Methods, e-Methods, links.lww.com/NXG/A245). RT4D6- erozygotes, this apparently produces no changes in myelin de- P2T rat Schwannoma cells were obtained from Professor Dies velopment and nerve function. Of interest, those members of the Meijer (University of Edinburgh, UK). HEK293 cells were family who also harbor a nonpathogenic very-low-frequency obtained from Sigma-Aldrich (St. Loius, MO). polymorphism (<0.001) in LITAF (LITAF T49M)10 developed a full-blown clinical pattern of inherited demyelinating peripheral Reporter activity assays polyneuropathy. Our data suggest that in these patients, a mild Schwann cells or HEK293 cells were transfected with the in- reduction in the function of the LITAF T49M protein can in- dicated constructs and then lyzed. Their luciferase activity was terfere with intracellular signaling and decrease the expression of determined with the Luciferase Assay System (Promega, Madison, an already transcriptionally deficient EGR2 P397H, hampering WI) using the manufacturer’s recommendations. The β-galacto- myelin development and nerve function. sidase activity (Beta-Glo Assay System; Promega) of a pRSV40- LacZ reporter cotransfected at 1:100 was used to normalize var- iations in cell number, viability, and transfection efficiency. Methods Immunofluorescence Family selection Immunofluorescence studies were performed on cells cul- We selected a white family from the southwest region of Spain tured in coverslips using standard procedures (see Methods, without consanguinity (figure 1). In this family, the proband e-Methods, links.lww.com/NXG/A245). patient and her father had an intermediate hereditary motor and sensory neuropathy. Chromatin immunoprecipitation assays The chromatin immunoprecipitation (ChIP) assay followed Next-generation sequencing and analyses a modified procedure of the method12 (see Methods, e-Methods, We extracted DNA from blood and then performed next- links.lww.com/NXG/A245). generation sequencing of a panel of 34 genes correlated with Charcot-Marie-Tooth disease (table e-1, links.lww.com/NXG/ DNA-protein binding assay A244). The regions under study were selected by hybridization The binding of EGR2 to DNA was performed using the DNA- including exons and adjacent intron regions (−8, +8) of the Protein Binding Assay Kit from Abcam (ab117139) (see panel genes. Clonal amplification and sequencing was per- Methods, e-Methods, links.lww.com/NXG/A245). formed using a paired-end strategy on the Illumina MiSeq platform. Sanger sequencing was accomplished for relevant Standard protocol approvals, registrations, detected changes and in silico predictive analyses for missense and patient consents mutations through the Alamut program. We screened variants’ Human DNA was obtained using protocols approved by our frequencies at the Inherited Neuropathy Variant Browser (hihg. institutional review board. The research was performed in ac- med.miami.edu/code/http/cmt/public_html/index.html#/). cordance with the Declaration of Helsinki. The Ethical

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 1 Segregation of EGR2 P397H and LITAF T49M mutations in the family

(A) Images obtained from the proband (III.3) showing atrophy of the posterior leg muscle compartment that generates lower limbs in inverted champagne bottle. In the upper limbs, hand muscles and thenar eminence were also atrophied. (B) Family pedigree. Ages at the moment of exploration are indicated in red. The arrow indicates the proband. LITAF neg, EGR neg: tested for mutations with negative results. (C) Chromatograms of members of the family. A red arrow indicates the position of the potentially mutated nucleotide. (D) Cartoon showing the localization of the newly identified mutation in the loop connecting zinc fingers 2 and 3 of the EGR2 transcription factor (arrow in the upper image). The alignment of the equivalent region of EGR2 from 7 different species shows that this is a highly evolutionary conserved sequence (lower image). LITAF = lipopolysaccharide-induced TNF-α factor.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Committee of the General Hospital of Alicante approved the proband’s genomic DNA by PCR and sequencing with the study. All adult participants provided informed consent for clin- Sanger method (III.3 in figure 1C). ical and genetic analysis. EGR2 (c.1190C>A; p.Pro397His) associates with Statistical analysis CMT disease only in a LITAF (c.146C>T; Values are presented as mean ± standard error among com- p.Thr49Met) genetic background parison groups, and statistical significance of group differences To determine whether the EGR2 P397H mutation is associated was estimated with the Student t test. A critical p value of less with the disease, we performed a cosegregation study of this than 0.05 was considered significant. Analysis was performed mutation with CMT symptoms in other members of the family. using GraphPad software (version 6.0). The proband’s father, aged 79 years (II.3 in figure 1B), referred instability, falls, and distal paresthesia for about 20 years. Our Data availability neurologic examination showed hand interosseous and distal lower The data that support the study findings are available from the limb atrophy with almost normal muscle strength. Reflexes were corresponding authors on reasonable request. decreased in the legs, and arched feet were also observed. Elec- trodiagnostic study demonstrated a chronic intermediate sensory and motor polyneuropathy, with predominance in the lower limbs Results (table 1). The CMTNS value was 9, indicating a mild disability. Clinical studies of the proband EGR2 (c.1190C>A; p.Pro397His) was found in heterozygosis The proband (III.3 in figure 1B) was a 54-year-old woman with in the father, suggesting an association of the mutation with the an intermediate hereditary motor and sensory neuropathy, with symptoms. However, he also expresses the LITAF T49M var- chronic axonal damage diagnosed at age 47 years in a neuro- iant, precluding the possibility of establishing an unequivocal physiologic study. Initial symptoms began in the early childhood association between the EGR P397H mutation and the CMT with progressive muscular atrophy and gait instability. At the symptoms. The proband’s mother (II.4 in figure 1, B and C) most recent examination in April 2018 (figure 1A), the distal does not have these mutations in EGR2 and LITAF genes. upper and lower limbs showed moderate-severe atrophy with associated weakness, arched feet, and Achilles tendon shortening. The proband’s brother (III.2 in figure 1B) was aged 45 years. Reflexes were decreased or abolished. All sensory modalities were His neurologic examination showed slight lower distal limb diminished. The CMT neuropathy score (CMTNS)13 value was atrophy, flat feet, and Achilles tendon retraction. Nerve con- 20. In the neurophysiologic study, motor nerve conduction ve- duction velocity studies revealed a selective motor axonal locities were slightly decreased in the median nerve, and no neuropathy in both tibial nerves without data of complete compound action potentials could be detected in the lower limbs polyneuropathy (table 1). Genetic studies showed the LITAF (peroneal and tibial nerves) (table 1). Sensory nerve conduction T49M variant (in heterozygosis) but no mutation in the velocities were decreased in the median and sural nerves, and EGR2 gene (figure 1C). The son (IV.1 in figure 1, B and C) action potentials were undetectable in the peroneal nerve. showed the LITAF T49M in heterozygosis but was asymp- tomatic and neurologically normal. The daughter (IV.2 in Genetic study of the proband figure 1, B and C) was also asymptomatic and did not show The proband (III.3 in figure 1B) underwent a detailed genetic any of these mutations. Next, we explored other members of study to identify the mutated gene that was putatively re- the family. Two of the proband’s aunts were likely affected, sponsible for the disease. After ruling out PMP22 duplication, but one of them (II.1 in figure 1B) died before the study, and 34 candidate genes (table e-1, links.lww.com/NXG/A244) the other (II.2 in figure 1B) did not want to participate in the associated with CMT disease were selected for next- study. The 55-year-old proband’s cousin (III.1 in figure 1, B generation sequencing. No previously described pathogenic and C) was also examined. He had never complained of mutations were found in any of these genes. However, the symptoms, and the neurologic examination was normal. patient shows a previously described variant of the LITAF Electrophysiology revealed an exclusively sensitive de- gene (c.146C>T; p.Thr49Met), considered to be a poly- myelinating polyneuropathy, probably due to the diabetes morphism, as it has been found in both symptomatic and mellitus type 1 background (table 1). Of interest, the genetic nonsymptomatic individuals7,10 (figure 1, B and C). This study revealed that he had the EGR2 P397H variant in het- variant is predicted to produce the change of threonine 49 by erozygosis but no mutation in LITAF. Thus far, our data a methionine in the LITAF polypeptide (LITAF T49M). Of showed that only those members of the family with the EGR2 interest, the proband also demonstrated a previously unde- P397H and LITAF T49M develop CMT symptoms. scribed missense mutation in EGR2 (c.1190C>A; p.Pro397- His) (figure 1, B and C). This nucleotide substitution is EGR2 P397H has a decreased predicted to produce a change of the proline 397 by a histidine transcriptional activity (EGR2 P397H) within the loop sequence that connects zinc Mutations in EGR2 cause distinct forms of demyelinating finger numbers 2 and 3 (figure 1D). The presence of both peripheral neuropathies including CMT1 type D and mutations in heterozygosis was confirmed by amplifying the Dejerine-Sottas neuropathy, among others.6 Many of these

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 1 Electrophysiologic findings of the proband (III-3), father (II-3), brother (III-2), and cousin (III-1)

Proband III-3 II-3 III-3 III-1

Motor nerve conduction

Median nerve

Distal latency (ms) (<4.2) 4.40 4.75 2.85 3.9

CMAP (mV) (>3.5) 4.9 3.5 5.6 6.5

MCV (m/s) (>48) 45.4 44.5 54.4 49.3

Ulnar nerve

Distal latency (ms) (<3.4) 3.05 2.1 2.65 3.3

CMAP (mV) (>2.8) 5.0 6.6 6.4 6.3

MCV (m/s) (>50) 47.8 59.4 Normal Normal

Peroneal nerve

Distal latency (ms) (<5.5) NR NR 4.1 Normal

CMAP (mV) (>2.5) NR NR 5.8 Normal

MCV (m/s) (>40) NR 14.8 57.1 Normal

Tibial nerve

Distal latency (ms) (<6.0) NR 8.05 3.65 5.0

CMAP (mV) (>2.9) NR 0.1 2.0 4.1

MCV (m/s) (>41) NR NR 30.9 41.2

Sensory nerve conduction

Median nerve

SNAP (μV) (>19) 5.8 3.7 39.6 Normal

SCV (m/s) (>47) 43.0 29.8 51.4 Decreased

Ulnar nerve

SNAP (μV) (>18) 4.2 5.8 24.8 Normal

SCV (m/s) (>44) 36.6 42.6 48.8 Normal

Peroneal nerve

SNAP (μV) (>3.5) NR 1.2 4.6 3.9

SCV (m/s) (>40) NR 26.7 52.9 36.5

Sural nerve

SNAP (μV) (>4.9) 4.8 0.43 6.5 5.4

SCV (m/s) (>41.3) 33.8 25.6 47.9 41.7

Abbreviations: CMAP = compound muscle action potential; HMSN = hereditary sensory and motor neuropathy; MCV = motor conduction velocity; NR = nonrecordable; SCV = sensory conduction velocity; SNAP = sensory nerve action potential. Abnormal values are written in bold. Precise data for the peroneal and median nerve for patient III-1 were not available. The nerve conduction velocities and amplitudes were predominantly more decreased in the lower than in the upper limbs, with peroneal and sural nerve severe injury in the proband and her father. Peroneal nerve motor and sensory conduction velocities were NR for the proband. Tibial nerve motor conduction velocity was NR for the proband (III-3) and her father (II-3). Proband III-3 had motor tibial nerve damaged with a prevailing axonal pattern. The proband’s cousin (patient III-1) showed a selective sensory peroneal and right median nerve conduction velocity decline without HMSN complete criteria. mutations map in 1 of the 3 zinc finger domains of the tran- finger 3 (figure 1D). Importantly, mutations in the equivalent scription factor. The consequence of these mutations is generally proline (P368) in nerve growth factor induced gene A (a related the loss of DNA binding capacity. Although not strictly located zinc finger protein) block its transcriptional activity.14 To learn within the zinc fingers, proline 397 is an evolutionary conserved how P397H change affects EGR2 function, we introduced this residue located within a loop that connects zinc finger 2 with zinc mutation in the construct pcDNA3.1 hEGR2. Seeking to

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 determine whether the mutation alters the stability of EGR2 the Schwann cells of the double heterozygotes. One possibility was polypeptide, HEK293 cells were transfected with wild-type and that this mutation had a loss of function effect by decreasing mutant constructs. As is shown in figure 2A, both proteins were protein stability. Of interest, during the coimmunoprecipitation expressed at similar levels, illustrating that the mutation does not studies performed in HEK cells (figure e-1, links.lww.com/NXG/ affect protein stability. Moreover, the EGR2 P397H mutant lo- A243), we noticed that the LITAF T49M GFP fusion protein was calized into the nucleus probing that the mutation does not consistently much less expressed compared with the LITAF-GFP change the subcellular localization of this transcription factor wild type. To learn whether the T49H mutation affects LITAF (figure 2B). Then, we evaluated the transcriptional activity of protein stability also in Schwann cells, we transfected cultured EGR2 P397H using an enhancer cloned upstream of the human Schwann cells with LITAF-GFP and LITAF T49M GFP con- P2 PMP22 promoter sequence and the pGL4 luciferase re- structs. As is shown in figure 3D, the number of cells expressing porter.15 As is shown in figure 2C, EGR2 wild type induced detectable levels of LITAF protein was notably lower when the luciferase activity by more than 28-fold over the basal activity mutant construct was transfected, suggesting decreased protein (pcDNA3.1 empty vector). By contrast, the EGR2 P397H mu- stability. To further substantiate this hypothesis, RT4D6 tant increased the activity by only 13-fold. Schwannoma cells were transfected with the LITAF T49M GFP and the LITAF-GFP constructs. After 48 hours, cells To better approach the in vivo situation in the nerves of our were lyzed, and protein extracts were immunoblotted with patients, we decided to explore the transcriptional activity of the a polyclonal anti-LITAF antibody. Anti-glyceraldehyde-3- EGR2 P397H mutant in transfected primary cultures of Schwann phosphate dehydrogenase immunoblot was used as a loading cells.Asisshowninfigure 2D, EGR P397H is normally expressed control. As is shown in figure 3E, the LITAF T49M GFP and localized and has decreased transcriptional activity, including protein was less abundant than LITAF-GFP also in these in Schwann cells as well. To determine whether the loss of cells. A similar result was obtained when the GFP tag was transcriptional activity is caused by decreased capacity of the removed from the constructs. Together, our results show EGR2 P397H to bind DNA, HEK293 cells were transfected with that the T49M mutation produces a decrease in the stability the wild-type and mutant EGR2 and nucleus isolated (see of the LITAF polypeptide that could interfere with NRG1- Methods). Nuclear extracts were allowed to bind a double- erbB signaling in Schwann cells. stranded biotinylated oligonucleotide with a tandem-repeated EGR2-binding consensus sequence (see Methods, e-Methods, links.lww.com/NXG/A245). DNA-protein complexes were Discussion captured in the assay microwell and quantified colorimetrically with an anti–EGR2-specific antibody (see Methods, e-Methods, CMT disease is a hereditary peripheral neuropathy caused links.lww.com/NXG/A245). As is shown in figure 3A, both wild- by distinct mutations in more than 40 different genes. One of type and mutant EGR2 showed similar binding capability DNA. these genes is EGR2, a pivotal master gene in the estab- To further confirm this result, we performed ChIP assays on the lishment of the myelin gene expression program by Schwann 4,6,12 PMP22 enhancer in transfected HEK293 cells (see Methods). As cells. Indeed, more than 30 missense mutations causing is highlighted in figure 3B, ChIP assay confirmed that the P397H peripheral neuropathy have been described in this tran- mutation does not change the capacity of EGR2 to bind DNA. scription factor (hihg.med.miami.edu/code/http/cmt/pub- lic_html/index.html#/). Most of these mutations are LITAF T49M protein has decreased stability located in the zinc finger region, a domain essential for the Initial studies postulated that LITAF itself might act as a tran- transcriptional activity of EGR2. Using next-generation se- scription factor.8,16 Indeed, recent studies have shown that quencing, we discovered a novel mutation of this domain in LITAF has a cysteine-rich region that coordinates zinc.17,18 We EGR2 (EGR2 P397H) in a family with CMT clinical explored whether LITAF could positively regulate the tran- symptoms. Using promoter activity assays, we found that scription of myelin genes. To this aim, HEK293 cells were this mutant has a decreased transcriptional activity com- cotransected with phPMP22enh-Luciferase and the plasmids pared with the wild type. Surprisingly, DNA binding and encoding for the human LITAF sequence. As shown in figure 3C, ChIP assays suggest that this mutant transcription factor LITAF was not able to induce the transcription of PMP22,nei- binds efficiently to myelin gene promoters. Of interest, ther increase the transcriptional activity of EGR2, ruling out a similar mutation in the zinc finger transcription factor a direct upregulation of EGR2 transcriptional activity by this nerve growth factor induced gene A (P369) that blocks protein. We also ruled out a direct physical interaction between transcriptional activity does not interfere with DNA bind- LITAF and EGR2 proteins by coimmunoprecipitation studies ing.14 Together, these data suggest that this proline residue is (figure e-1, links.lww.com/NXG/A243). LITAF has been shown involved in the recruitment of other proteins such as coac- to participate in the recruitment of endosomal sorting complex tivators, but not in the binding of EGR2 to DNA. Experi- required for transport components to endosomal membranes ments to address this point are currently being performed in regulating ErbB receptor trafficking and consequently NRG1 our laboratory. signaling in Schwann cells.9 Because NRG1 signaling is pivotal for EGR2 induction and Schwann cell differentiation,19 we hypoth- We found that this family also expressed a mutation in LITAF esized that LITAF T49M may interfere with EGR2 expression in (LITAF T49M). Although other missense mutations in this

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 2 EGR2 P397H shows a decreased transcriptional activity in both heterologous systems and Schwann cells

(A) Western blot of HEK293 cells transfected with an empty vector (pcDNA3.1), pcDNA3.1 hEGR2 (EGR2), and pcDNA3.1 hEGR2 P397H (EGR2 P397H). GAPDH was used as a loading control. No change in the stability of the mutant protein is observed. (B) Immunofluorescence with anti-EGR2 antibody of HEK293 cotrans- fected cells with the different EGR2 constructs and GFP demonstrate that both wild-type and mutant proteins localize properly in the nu- cleus (labeled with Hoechst staining). Anti-GFP immunofluorescence highlights those cells that have been transfected. Bars 25 μm. (C) P397H decreases the transcriptional activity of EGR2. HEK293 cells were transfected with phPMP22enh-Luciferase and an empty vector (pcDNA3), EGR2 or EGR2 P397H and luciferase activity determined. Results were normalized against the activity of the pRSV40-LacZ vector. Data are given as mean ± standard error (SE) and analyzed with the t test (2 sided). ****p < 0.0001. (D) Immunofluorescence with anti- EGR2 antibody of Schwann cells cotransfected with the different EGR2 constructs and GFP showing both wild-type and mutant proteins localized in the nucleus. Anti-GFP immunoflu- orescence highlights transfected cells. Bars 25 μm. (E) As in HEK293 cells, the transcriptional activity of EGR2 P397H also decreases in Schwann cells. Data are given as mean ± SE and analyzed with the t test (2 sided). ****p < 0.0001. GAPDH = glyceraldehyde-3-phosphate dehydrogenase; GFP = green fluorescent pro- tein; LITAF = lipopolysaccharide-induced TNF-α factor.

gene cause CMT type 1C disease, this particular variant has been that both mutations are necessary to develop the clinical considered a polymorphism, as it was found in asymptomatic symptoms, we cannot completely rule out that an in- members of other pedigrees.10 Indeed, our analysis confirms that complete penetrance of the EGR P397H mutation could at least 2 members of the family who present this mutation (III.2 explain our findings. Genetic and clinical characterization of and IV.1) have no CMT compatible clinical symptoms and are other lineages will help to clarify this point. normal in the neurologic exploration. Initially, we reasoned that a physical interaction between Surprisingly, and despite of the decreased transcriptional LITAF and EGR2 proteins could explain the genetic in- activity of the EGR2 P397H, only those members of the teraction. However, we could not detect any binding of LITAF family who have both mutations at the same time developed to EGR2, neither a positive effect of LITAF on the transcrip- CMT symptoms (members II.3 and III.3). Indeed, the tional activity of EGR2. cousin of the proband (member III.1), who has only the EGR2 P397H mutation, and is of similar age, does not show It has been shown that the loss of function of LITAF pro- CMT symptoms. Although this provides strong evidence duces deregulation in the NRG1-erbB signaling in Schwann

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 Figure 3 DNA binding of EGR2P397H, transcriptional effects of LITAF, and stability of LITAF T49M

(A) EGR2 P397H binds normally to DNA in vitro. HEK293 cells were transfected with the constructs pcDNA3.1-EGR2 and pcDNA3.1-EGR2 P397H and nuclear extracts obtained. Biotinylated oligonucleotides with the EGR2-binding sequence in tandem (3 repeats) were incubated with nuclear extracts. Anti–EGR2 polyclonal antibody was used to quantify colorimetrically the binding of EGR2 to the oligonucleotide (see Methods). Nuclear extracts of nontransfected and GFP-transfected cells were used as controls. Data are given as mean ± standard error (SE) and analyzed with the t test (2 sided). (n = 7). (B) EGR2 P397H binds normally to DNA in vivo. HEK293 cells were transfected with pcDNA3.1-EGR2 and pcDNA3.1-EGR2 P397H and cross-linked with PFA. Chromatin was purified and immunoprecipitated with anti-EGR2 antibody or a nonspecific IgG (chromatin immunoprecip- itation grade). qPCR was performed with specific primers for the enhancer region of PMP22. As shown, the recovery of the PMP22 enhancer region in the immunoprecipitate was similar in both cases. No relevant recovery was obtained with the nonspecific IgG. Data are given as mean ± SE and analyzed with the paired t test (2 sided). *p <0.05;**p < 0.01 (n = 10). (C) LITAF has no PMP22 transcriptional activity and does not increase EGR2 transcriptional activity. HEK293 cells were transfected with phPMP22enh-Luciferase and an empty vector (pcDNA3), LITAF, EGR2, or LITAF + EGR2 and luciferase activity determined. Results were normalized against the activity of the pRSV40-LacZ vector. Data are given as mean ± SE and analyzed with the t test (2 sided). A slight tendency to decrease the basal activity of the reporter by the LITAF construct was observed. *p <0.05.(D) LITAF T49M has decreased stability. Cultured Schwann cells were transfectedwithLITAF-GFPandLITAFT49MGFPconstructs.Thenumberofcells expressing detectable levels of LITAF protein was notably lower with the mutant construct. Bar 25 μm. Quantification of immunofluorescence showing the percentage of GFP positive per frame (n = 30). Data are given as mean ± SE and analyzed with the t test (2 sided). (E) Immunoblot analysis supports a decreased stability for the LITAF T49M polypeptide. RT4D6 cells were transfected with the LITAF T49M GFP and the LITAF-GFP constructs, lysed, and protein extracts immunoblotted with a polyclonal anti-LITAF antibody. Anti-GAPDH immunoblot was used as a loading control. A similar result was obtained when the GFP tag was removed. The densitometry of 3 different experiments is shown. Data are given as mean ± SE and analyzed with the t test (2 sided). *p < 0.05; **p < 0.01. (F) Working model: a reduction in the function of the LITAF T49M protein decreases the expression of an already transcriptionally deficient EGR2 P397H hampering myelin development and nerve function. GAPDH = glyceraldehyde-3- phosphate dehydrogenase; GFP = green fluorescent protein; LITAF = lipopolysaccharide-induced TNF-α factor; PFA = paraformaldehyde; qPCR: quantitative polymerase chain reaction.

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG cells.9 Of interest, we observed that the steady-state levels of the LITAF T49M mutant protein (in both heterologous Appendix (continued) systems and in Schwann cells) are consistently much lower Name Location Contribution than the wild-type form, suggesting a loss of function mecha- nism. Because adequate NRG1-erbB signaling is pivotal for Sergio ISABIAL (FISABIO), Hospital Major role in the Velasco- General Universitario de acquisition of data and EGR2 expression and myelin development, it is conceivable Aviles, MSc Alicante and Instituto de analyzed and interpreted that although a partial decrease in the transcriptional activity of Neurociencias de the data Alicante UMH-CSIC, EGR2 P397H or a partial loss of LITAF function does not Spain individually interfere with myelin development, in the com- Angeles ISABIAL (FISABIO), Hospital Major role in the pound heterozygotes, the deregulated NRG1-erbB signaling Casillas- General Universitario de acquisition of data (consequence of a partial loss of LITAF function) will de- Bajo, BSc Alicante and Instituto de fi Neurociencias de Alicante ciently induce an already transcriptional defective EGR2, in- UMH-CSIC, Spain terfering with myelin development and peripheral nervous system function. Juan Salas- Hospital Marina Salud, Major role in the Felipe, MD Denia, Spain acquisition of data

Acknowledgment Alexandre Hospital IMED Levante, Major role in the Garc´ıa- Benidorm, Spain acquisition of data The authors thank the participating relatives of the patient for Escriv´a, MD their cooperation throughout this study. They thank Dr. Cruz Carmen ISABIAL (FISABIO), Hospital Designed and Morenilla-Palao (Instituto de Neurociencias UMH-CSIC, D´ıaz-Mar´ın, General Universitario de conceptualized the study; Spain) for advice in ChIP and other molecular biology MD, PhD Alicante, Spain analyzed and interpreted experiments. They thank Dr. Philip Woodman and Dr. Lydia the data; and drafted the manuscript Wunderley (University of Manchester, UK) for providing for intellectual pGFP-hLITAF and pGFP-hLITAF T49M plasmids, John content Svaren (University of Wisconsin) for pPMP22enh-Luciferase, Hugo ISABIAL (FISABIO), Hospital Designed and and Dies Meijer (University of Edinburgh, UK) for RT4D6 Cabedo, General Universitario de conceptualized the study; MD, PhD Alicante and Instituto de analyzed and Schwannoma cells. They also thank Pedro Morenilla-Ayala Neurociencias de Alicante interpreted the data; for technical assistance. UMH-CSIC, Spain and drafted the manuscript for intellectual content Study funding This work has been funded by grants BFU2016-75864R (MINECO Spanish Ministry for Economy and Competi- tiveness), PROMETEO 2018/114 (Conselleria Educaci´o References Generalitat Valenciana), and UGP-15-211 (FISABIO) to 1. Rossor AM, Polke JM, Houlden H, Reilly MM. Clinical implications of genetic H. Cabedo. advances in Charcot-Marie-Tooth disease. Nat Rev Neurol 2013;9:562–571. 2. Berciano J, Sevilla T, Casasnovas C, et al. Guidelines for molecular diagnosis of Charcot-Marie-Tooth disease [in Spanish]. Neurologia 2012;27:169–178. Disclosure 3. Gonzaga-Jauregui C, Harel T, Gambin T, et al. Exome sequence analysis suggests that genetic burden contributes to phenotypic variability and complex neuropathy. Cell Disclosures available: Neurology.org/NG. Rep 2015;12:1169–1183. 4. Nagarajan R, Svaren J, Le N, Araki T, Watson M, Milbrandt J. EGR2 mutations in inherited neuropathies dominant-negatively inhibit myelin gene expression. Neuron Publication history 2001;30:355–368. Received by Neurology: Genetics May 8, 2019. Accepted in final form 5. Srinivasan R, Sun G, Keles S, et al. Genome-wide analysis of EGR2/SOX10 binding in – December 30, 2019. myelinating peripheral nerve. Nucleic Acids Res 2012;40:6449 6460. 6. Warner LE, Svaren J, Milbrandt J, Lupski JR. Functional consequences of mutations in the early growth response 2 gene (EGR2) correlate with severity of human myeli- nopathies. Hum Mol Genet 1999;8:1245–1251. Appendix Author 7. Saifi GM, Szigeti K, Wiszniewski W, et al. SIMPLE mutations in Charcot-Marie- Tooth disease and the potential role of its protein product in protein degradation. Name Location Contribution Hum Mutat 2005;25:372–383. 8. Lee SM, Olzmann JA, Chin LS, Li L. Mutations associated with Charcot-Marie- Maria ISABIAL (FISABIO), Hospital Major role in the Tooth disease cause SIMPLE protein mislocalization and degradation by the Empar General Universitario de acquisition of data; proteasome and aggresome-autophagy pathways. J Cell Sci 2011;124: Blanco- Alicante, Spain designed and 3319–3331. Canto,´ MD conceptualized the study; 9. Lee SM, Chin LS, Li L. Charcot-Marie-Tooth disease-linked protein SIMPLE analyzed and interpreted functions with the ESCRT machinery in endosomal trafficking. J Cell Biol 2012; the data; and drafted the 199:799–816. manuscript for 10. Beauvais K, Furby A, Latour P. Clinical, electrophysiological and molecular genetic intellectual content studies in a family with X-linked dominant Charcot-Marie-Tooth neuropathy pre- senting a novel mutation in GJB1 promoter and a rare polymorphism in LITAF/ Nikiben ISABIAL (FISABIO), Hospital Major role in the SIMPLE. Neuromuscul Disord 2006;16:14–18. Patel, MSc General Universitario de acquisition of data; 11. Brockes JP, Fields KL, Raff MC. Studies on cultured rat Schwann cells. I. Establish- Alicante and Instituto de analyzed and interpreted ment of purified populations from cultures of peripheral nerve. Brain Res 1979;165: Neurociencias de Alicante the data; and drafted the 105–118. UMH-CSIC, Spain manuscript for 12. Jang SW, LeBlanc SE, Roopra A, Wrabetz L, Svaren J. In vivo detection of Egr2 intellectual content binding to target genes during peripheral nerve myelination. J Neurochem 2006;98: 1678–1687.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 13. Shy ME, Blake J, Krajewski K, et al. Reliability and validity of the CMT neuropathy cloning, sequencing, characterization, and chromosomal assignment. Proc Natl Acad score as a measure of disability. Neurology 2005;64:1209–1214. Sci USA 1999;96:4518–4523. 14. Wilson TE, Day ML, Pexton T, Padgett KA, Johnston M, Milbrandt J. In vivo 17. Ho AK, Wagstaff JL, Manna PT, et al. The topology, structure and PE interac- mutational analysis of the NGFI-A zinc fingers. J Biol Chem 1992;267: tion of LITAF underpin a Charcot-Marie-Tooth disease type 1C. BMC Biol 2016;14:109. 3718–3724. 18. Qin W, Wunderley L, Barrett AL, High S, Woodman PG. The Charcot Marie Tooth 15. Jones EA, Lopez-Anido C, Srinivasan R, et al. Regulation of the PMP22 gene through disease protein LITAF is a zinc-binding monotopic membrane protein. Biochem J an intronic enhancer. J Neurosci 2011;31:4242–4250. 2016;473:3965–3978. 16. Myokai F, Takashiba S, Lebo R, Amar S. A novel lipopolysaccharide-induced tran- 19. Arthur-Farraj P, Wanek K, Hantke J, et al. Mouse schwann cells need both NRG1 and scription factor regulating tumor necrosis factor alpha gene expression: molecular cyclic AMP to myelinate. Glia 2011;59:720–733.

10 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG ARTICLE OPEN ACCESS Molecular diagnosis of muscular diseases in outpatient clinics A Canadian perspective

Fanny Thuriot, MSc, Elaine Gravel, MSc, Caroline Buote, MSc, Marianne Doyon, MD, Elvy Lapointe, MSc, Correspondence Lydia Marcoux, MSc, Sandrine Larue, MD, Am´elie Nadeau, MD, S´ebastien Ch´enier, MD, Paula J. Waters, PhD, Fanny Thuriot [email protected] Pierre-´Etienne Jacques, PhD, Serge Gravel, PhD, and S´ebastien L´evesque, MD, PhD

Neurol Genet 2020;6:e408. doi:10.1212/NXG.0000000000000408 Abstract Objective To evaluate the diagnostic yield of an 89-gene panel in a large cohort of patients with suspected muscle disorders and to compare the diagnostic yield of gene panel and exome sequencing approaches.

Methods We tested 1,236 patients from outpatient clinics across Canada using a gene panel and per- formed exome sequencing for 46 other patients with sequential analysis of 89 genes followed by all mendelian genes. Sequencing and analysis were performed in patients with muscle weakness or symptoms suggestive of a muscle disorder and showing at least 1 supporting clinical laboratory.

Results We identified a molecular diagnosis in 187 (15.1%) of the 1,236 patients tested with the 89- gene panel. Diagnoses were distributed across 40 different genes, but 6 (DMD, RYR1, CAPN3, PYGM, DYSF, and FKRP) explained about half of all cases. Cardiac anomalies, positive family history, age <60 years, and creatine kinase >1,000 IU/L were all associated with increased diagnostic yield. Exome sequencing identified a diagnosis in 10 (21.7%) of the 46 patients tested. Among these, 3 were attributed to genes not included in the 89-gene panel. Despite differences in median coverage, only 1 of the 187 diagnoses that were identified on gene panel in the 1,236 patients could have been potentially missed if exome sequencing had been per- formed instead.

Conclusions Our study supports the use of gene panel testing in patients with suspected muscle disorders from outpatient clinics. It also shows that exome sequencing has a low risk of missing diagnoses compared with gene panel, while potentially increasing the diagnostic yield of patients with muscle disorders.

From the Department of Pediatrics (F.T., E.G., C.B., M.D., L.M., A.N., S.C., P.J.W., S.G., S. L´evesque), Universit´e de Sherbrooke; Sherbrooke Genomic Medicine (F.T., E.G., C.B., S.G., S. L´evesque); RNomic’s Platform (E.L.), Universit´e de Sherbrooke; Department of Neurology (S. Larue), Notre-Dame Hospital, Universit´e de Montr´eal; Department of Biology (P.-´E.J.), Universit´e de Sherbrooke; and Department of Computer Sciences (P.-´E.J.), Universit´e de Sherbrooke, Quebec, Canada.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary CK = creatine kinase; LGMD = limb-girdle muscular dystrophy; VUS = variants of uncertain significance.

Inherited muscle disorders form a heterogeneous group specific sections for muscle phenotype, respiratory symp- characterized by different patterns of muscle weakness. Some toms, cardiac anomalies, and clinical laboratory results. Age have a unique phenotype allowing a diagnostic with single- at onset and previous genetic testing were not systematically gene testing. Others, like the limb-girdle weakness pattern, do mentioned. Furthermore, complete muscle biopsy reports not have a specific gestalt and are associated with numerous were not available to us. Respiratory and cardiac phenotypes genes. Muscle biopsy has long been considered the gold were described in 83.4% and 71.0% of patients, respectively. standard test but is not without limitations. Indeed, it is par- Cardiac anomalies included hypertrophic or dilated cardio- ticularly invasive, and specific antibodies are not available for myopathies and arrhythmias for 82.2% of patients and were all proteins encoded by causal genes.1 Therefore, genetic unspecified for the remaining. CK, EMG, muscle biopsy, and testing is increasingly used and can be obtained in a non- MRI results were available for 89.5%, 81.6%, 32.7%, and invasive manner. Across different countries such as Canada, 11.7% of patients, respectively. In addition, 46 patients were the United States, China, Korea, Germany, the United recruited using the above-mentioned criteria from neuro- Kingdom, Egypt, Poland, Australia, and Japan, gene panel muscular and genetics outpatient clinics at Centre Hospi- – sequencing has a yield varying from 16% to 65%,2 7 talier Universitaire de Sherbrooke. There was no overlap depending on subgroups of patients’ selection, whereas with the 1,236 patients above. Fifty-four percent had single – exome sequencing has a yield in between 13% and 69%8 15 in gene or panel testing prior performing exome sequencing. different settings. Its superiority over gene panel, in di- DNA was than analyzed by exome sequencing on a research agnosing common etiologies, remains to be quantified. basis, following genetic counseling. Because patients come from outpatient clinics, follow-up was made by treating Reaching a molecular diagnosis is becoming increasingly im- physicians. portant for patients management, including participation in clinical trials and treatment eligibility, such as enzyme re- Standard protocol approvals, registrations, placement in Pompe disease.16 and patient consents The study was approved by the institutional ethics review Here, we report on the diagnostic yield of an 89-gene panel in board at Universit´e de Sherbrooke (project ##MP-31-2013- a large cohort of Canadian patients with suspected muscle 533, 12-208). All participants (or their legal guardians) who disorders from outpatient clinics, and we compare the di- underwent exome sequencing provided participants’ written agnostic yield of gene panel and exome sequencing approaches consent. in a single-center cohort. Gene panel sequencing We designed a panel of 89 genes targeting diverse patterns of Methods muscle weakness that could be encountered in outpatient clinics and covering the following groups of disorders: limb- Recruitment of patients girdle muscular dystrophies (LGMDs), congenital muscular We analyzed DNA samples of 1,236 patients (201 children dystrophies, congenital myasthenic syndromes, nemaline and 1,035 adults; 574 females and 662 males) seen by 187 myopathy, myofibrillar myopathy, centronuclear myopathy, physicians in outpatient clinics (general neurology, special- collagen VI–related myopathies, inclusion myopathies, met- ized neuromuscular, genetics, physiatry, and general practice) abolic myopathies, rigid spine syndromes, and scapuloper- at 61 locations across Canada (figure e-1, links.lww.com/ oneal syndromes. The list of the 89 genes is provided in table NXG/A233). DNA samples were extracted from blood e-1 (links.lww.com/NXG/A234) and supplementary file e-1 samples. The clinical gene panel test was performed in the (links.lww.com/NXG/A240). For each patient, DNA librar- laboratory of Sherbrooke Genomic Medicine (a not-for-profit ies were prepared following a standard protocol (Kapa Bio- organization), and the cost of the test was covered by a special systems, Roche, MA), followed by target enrichment (Seq program with financial support from Sanofi Genzyme. To be Cap EZ-Custom) and sequenced on a MiSeq (Illumina, San eligible, patients were required to show any type of muscle Diego, CA) or a NextSeq (Illumina) with a 150-bp paired-end weakness or symptoms suggestive of muscle involvement protocol. A total of 383 and 853 patients were sequenced on (i.e., myalgia, rhabdomyolysis, exercise intolerance, and un- MiSeq and NextSeq, respectively. explained respiratory insufficiency), at least 1 abnormal lab- oratory finding suggestive of muscle involvement (plasma Exome sequencing creatine kinase [CK], EMG, muscle biopsy, or MRI), and no Exome sequencing was performed as previously described.8 reported diagnosis. Demographics and clinical information Briefly, it was performed at the McGill University and were obtained from the laboratory requisition, which included G´enome Qu´ebec Innovation Centre (Montreal, Canada) or

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Fulgent (Temple City, CA). DNA libraries were prepared for Results each patient (TruSeq; Illumina), followed by target enrich- ment (Agilent SureSelect All Exon kit v4 or v5 or Illumina Diagnoses identified by the 89-gene panel in Truseq Exome) and sequenced on a HiSeq 2000 (Illumina) outpatient clinics with a 100-bp paired-end protocol or HiSeq 4000 (Illumina) Samples from a total of 1,236 patients with a suspected muscle with a 150-bp paired-end protocol. disorder were analyzed by gene panel. Adults and children younger than 18 years accounted for 83.7% and 16.3%, re- spectively. A third of adult patients were aged 60 years and Bioinformatics and statistical analysis older. Ethnic background was reported to be European non- We analyzed the sequencing data using a Linux-based bio- Finnish in 67.5%, when data were available. Other ethnicities informatics pipeline based on the one developed by the McGill included South and East Asians (14.7%), Mixed Ethnicities University and G´enome Qu´ebec Innovation Centre (bitbucket. (5.1%), Middle East (3.5%), Native Americans and Canadians org/mugqic/mugqic_pipelines) as previously described.6 fl 17 (3.4%), African Americans (3.2%), Hispanic (1.7%), Ashke- Brie y, (1) raw reads were trimmed using Trimmomatic nazi Jewish (0.5%), and European Finnish (0.4%). Patients (version 0.32); (2) sequence alignment was performed with 18 who were recruited from specialized neuromuscular out- Burrows-Wheeler Aligner (version 0.7.10); (3) genetic var- patient clinics represented 61.7% of the total cohort, whereas iants (single nucleotide polymorphisms and indels) were called 19 other general neurology, clinical and biochemical genetics, with the HaplotypeCaller using the Genome Analysis Toolkit and other clinics accounted for 23.9%, 10.4%, and 4.0%, re- (version 3.2.2) with prior local realignment, base recalibration, spectively. More than half of our cohort presented with and removal of polymerase chain reaction (PCR) duplicates a limb-girdle weakness, in both children and adults (figure 1). using Picard (version 1.123, broad institute.github.io/picard/); ff 20 (4) gene annotation was performed with SnpE /SnpSift A total of 187 (15.1%) patients had a diagnosis identified (versions 3.6 and 4.2, including SIFT, Polyphen2, and Muta- (figure 2 and table e-2, links.lww.com/NXG/A235), with tionTaster predictions) with an additional in-house script to 21 a diagnostic rate of 22.4% in children and 13.7% in adult annotate variants present in the ClinVar database; and (5) patients. A higher diagnostic rate (27.8%) was observed a filtering process removed variants outside targeted sequences, 22 within males of the pediatric cohort due to DMD, a gene with population frequency >1% (dbSNP 138 and ExAC 0.3 ) causing an X-linked disorder. Otherwise, there was not any and genotype quality less than Q30. Coverage depth was cal- 23 gender bias. A likely carrier status of a recessive disorder was culated using BED Tools. Filtered variant lists obtained from identified in 9.9%. A potential diagnosis was suspected in the bioinformatics pipeline were then interpreted with an in- 16.4% of the cases (table 1). In particular, 66 patients had 2 house script and manual revision. For both gene panel and variants in a gene associated with an autosomal recessive exome sequencing, deletion and duplication analysis were disorder compatible with the patient’s phenotype, including performed using the CoNVaDING software24 and manual fi 33 patients with 1 pathogenic variant and 1 VUS. A biopsy was review of binary alignment map les before quantitative PCR performed in 18 of 33 patients, of which 17 were abnormal. confirmation using Taqman Copy Number Assay (Thermo fi fi Among these 17 abnormal biopsies, 14 showed nonspeci c Fisher Scienti c, Montreal, Canada). Variants were revised findings. Two patients harboring NEB variants showed manually and were reported according to the American Col- 25 nemaline bodies on muscle biopsy, which supported a di- lege of Medical Genetics and Genomics guidelines. For agnosis of nemaline myopathy, but parents were not tested to exome sequencing, a subset of 89 genes (included in the gene document whether variants were inherited in trans (patients panel) was analyzed before the whole-exome data. Variants of 336 and 372, see table e-3, links.lww.com/NXG/A236). One uncertain significance (VUS) were reported only if related to ’ fi patient harboring SGCA variants showed absent alpha- the patient s phenotype. Sanger sequencing con rmation was sarcoglycan on muscle biopsy, which supported a diagnosis only performed if any of the following criteria were not ful- fi of autosomal recessive LGMD type 2D, but parents were not lled: minimum genotype quality >Q40, quality score >500, tested to document whether variants were inherited in trans strand bias score <60, and heterozygous read ratio >60/40. (patient 356, see table e-3). Overall, information was in- sufficient to confirm the diagnosis in these 66 patients. Pa- Odds ratios (figure 4) were calculated for clinical criteria by fi χ2 rental testing and additional speci c immunochemistry, when using standard statistics ( test). When data were missing on applicable, were therefore recommended. a patient for a specific clinical criterion, we removed the pa- tient from the analysis and calculated among patients with The 187 confirmed diagnoses were distributed across 40 different complete data. genes, but 6 explained approximately 50% of all cases (figure 3C and table e-2, links.lww.com/NXG/A235). DMD was the most Data availability common etiology (figure 3). Among the patients with causal Anonymized data will be shared by request from any qualified variants in DMD, all the 17 adult patients had Becker muscular investigator. When not possible, given the risk to identify rare dystrophy, and 4/11 pediatric patients had Duchenne muscular patients, additional aggregate data in table form will be pro- dystrophy. The latter were all due to a single nucleotide variation duced to address specific questions. (table e-5, links.lww.com/NXG/A238). Half of patients with

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Figure 1 Clinical presentation of the pediatric (A) (n = 201) and adult (B) (n = 1,035) cohorts

Becker muscular dystrophy had an exon deletion. Besides DMD, 24.9%. Among patients with diagnoses, CK value or cardiac the genes most commonly found to be responsible for muscular information was available on 179 patients, of whom only disease in our cohort were RYR1, CAPN3, PYGM, DYSF,and 10.7% had normal CK value and no cardiac anomaly. There FKRP (figure 3). Among the 187 diagnoses, we identified 248 was no significant difference in diagnostic yield between causal variants: 8.1% were large deletions (exon or whole gene muscle weakness patterns. Finally, although it did not reach deletion), 23.0% small indels, and 68.9% substitutions. Twenty- statistical significance, the medical specialty of the ordering six percent of the causal variants are new variants not previously physician showed some differences in the diagnostic rate: reported (64 variants). Of the 20 large deletions, the majority 17.2% (22/128) for geneticists, 17.0% (130/763) for neuro- were hemizygous DMD deletions. We also observed 1 hetero- muscular specialists, and 9.5% (28/295) for general neurol- zygous KLHL41 whole gene deletion, 1 heterozygous GAA ogists. Other medical specialties accounted for only a small whole gene deletion, 2 heterozygous PYGM exon deletions, 1 proportion of patients tested (50/1,236), but 7 diagnoses heterozygous DYSF exon deletion, and 1 heterozygous ANO5 were made among those 50 patients. exon deletion (table e-2, links.lww.com/NXG/A235). Diagnostic yield of exome sequencing Clinical characteristics influencing the compared with gene panel diagnostic yield A total of 46 patients from 1 center were recruited for exome Decreasing diagnostic yield was observed with increasing age, sequencing and analysis of 3,857 mendelian genes (list of ranging from 22.4% in the 0–17-year-old group, 16.4% be- genes provided in supplementary files e-2 and e-3, links.lww. tween 18 and 59 year olds, and dropping to 9.9% in patients com/NXG/A241, links.lww.com/NXG/A242). We first an- aged 60 years and older. Beside age <60 years, which was alyzed the 89 genes included in the gene panel and identified significantly associated with a higher diagnostic yield, other a diagnosis in 7/46 patients (15.2%). Three additional di- clinical criteria were investigated for their impact on the agnoses were attributed to genes not included in the muscle probability of identifying a molecular diagnosis on the gene panel (figure 4). Notably, the probability was significantly higher in patients having a known cardiac anomaly, a positive family history, or showing elevated CK, and in particular those Table 1 Potential diagnoses of our cohort of 1,236 patients with CK > 1,000 IU/L, who showed a diagnostic yield of No. of % of total Potential diagnoses patients patients

Rare VUS in a gene associated with AD 124 10.0 Figure 2 Proportion of the confirmed diagnoses (n = 187) disorder and potential diagnoses (n = 202) among our One pathogenic variant and 1 VUS in 33 2.7 1,236 patients a gene associated with AR disorder

Two heterozygous VUS in a gene 27 2.2 associated with AR disorder

One homozygous VUS in a gene 60.5 associated with AR disorder

Rare hemizygous VUS in a gene 12 1.0 associated with XL disorder

Total 202 16.4

Abbreviations: AD = autosomal dominant; AR = autosomal recessive; VUS = variants of uncertain significance; XL = X-linked.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 3 Genes that accounted for the most diagnoses in (A) the pediatric cohort (n = 45), (B) the adult cohort (n = 142), and (C) the whole cohort (n = 187)

disorders 89-gene panel: 1 gene causing metabolic myo- ANO5 along with a variant of unknown significance, but in- pathies (HADHA), 1 gene causing a mixed nerve and muscle formation was insufficient to confirm this potential diagnosis. pathology (MYH14), and 1 gene associated with oculophar- Overall, the diagnostic rate of exome was increased to 21.7% yngeal muscular dystrophy (PABPN1). In addition, we compared with 15.2% in the (virtual) gene panel for this single identified a homozygous pathogenic variant in ABCA1 center cohort (patients’ details in table e-4). (Tangier disease), but it could not fully account for the patient’s symptoms (table e-4, links.lww.com/NXG/A237). Although use of exome sequencing could potentially increase Another patient was found to harbor a pathogenic variant in the diagnostic rate by analyzing more genes, lower base cov- erage could lead to missed diagnoses in the 89 genes included in the muscle disorder gene panel. To estimate the proportion of potentially missed diagnoses, we compared the base cov- Figure 4 Odds ratios of the different clinical criteria erage obtained for these 89 genes from our exome sequencing cohort with the gene panel cohort. On average, the median coverage was 179x (range: 109x–333x) among the 46 patients analyzed by exome sequencing in comparison to 555x (MiSeq, range: 244x–992x) and 1654x (NextSeq, range: 674x–3730x), in 1,236 patients investigated by gene panel according to the sequencing platform used. Among the 3,857 mendelian genes analyzed on exome, 99.61% of bases had adequate coverage as defined by ≥10x. With respect to the 89 genes selected for the muscle disorder gene panel, 99.92% of bases were covered at ≥10x on exome sequencing, similar to what was observed with gene panel on both sequencing Age is separated in 2 categories (<60 and ≥60 years). Diagnostic yield is platforms (99.92% with MiSeq and 99.95% with NextSeq). higher in patients aged <60 years. Some values could not be included be- Exon 1 of SELENON (SEPN1) showed poor coverage in both cause the results were not available for all patients (familial history = 938/ 1,236, creatine kinase [CK] = 1,106/1,236, quantitative CK = 642/1,236, EMG = exome and gene panel (irrespective of platform used) in all 1,008/1,236, muscle biopsy = 404/1,236, MRI = 146/1,236, respiratory in- sufficiency = 1,031/1,236, cardiac anomaly = 878/1,236). Error bars represent samples owing to extremely high GC content (86.89%). This 95% confidence interval. exon accounts for ;0.06% of bases of the 89 muscle disorder genes selected. The small difference in base coverage between

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 exome and gene panel was attributed to some specific exons represents up to 16.4% of patients, was not readily available to (mostly GC-rich exons 1) showing recurrent borderline to us. In particular, follow-up molecular testing in other family poor coverage in a minority of patients’ samples (ranging members and detailed muscle biopsy reports could have en- from 2.7% to 36.1%) on exome sequencing: B3GALNT2 abled the reclassification of some VUS as likely pathogenic and (exon 1), ISPD (exon 1), LMOD3 (exon 1), PLEC (exon 1), led to additional confirmed diagnoses. Thus, the overall di- SEPN1 (exon 3), SGCB (exon 1), TNNT1 (exons 4 and 5), agnostic yield could be as high as 31.5%, but the 15.1% certainly and VCP (exon 1). Likely pathogenic or pathogenic variants represents a more conservative estimate. Of note, a recent large (substitutions and indels) have been previously reported in sequencing effort in the United States involving more than the ClinVar database for 4 of these 8 exons. However, none of 4,000 patients with a suspected LGMD observed a 27% di- the 228 causal substitutions and indels found among the 187 agnostic yield with their 35-gene panel.7 The authors included, diagnoses on gene panel localized to these exons, and thus, no among the confirmed diagnoses, patients who were compound diagnosis would have been missed. In addition, 14 of the 20 heterozygotes for a variant of uncertain significance and a likely large deletions identified among the 187 diagnoses would or pathogenic variant in recessive genes. This contributed to probably have been correctly identified by exome sequencing, 5.5% of the 27% yield. Similarly, the yield of the present study as they are hemizygous or homozygous. However, the het- would have been 17.8% if we had included such patients. An- erozygous exon deletion involving the last 2 exons of ANO5 other factor that could have contributed to this relatively lower could potentially have been missed, as some patients showed diagnostic yield is the significant proportion of patients older base coverage below 50x, which is associated with decreased than 60 years (a third of adult patients). As observed in our performance of the CoNVaDING algorithm.24 Notably, this study, age >60 years is associated with a lower diagnostic yield deletion has been identified successfully in another patient on (9.9%). This could be due to an increased contribution of exome sequencing (patient EX11, table e-4, links.lww.com/ nongenetic disorders in that group. NXG/A237). The other 5 deletions were all in regions over 100X. At worst, among the patients investigated by gene Similarly to previous studies, we observed a high genetic – panel, exome sequencing could have potentially missed 1/248 heterogeneity among diagnoses.29 32 Of interest, the genes causal variants (0.4%) or 1 of 187 diagnoses (0.5%). most commonly represented among the confirmed diagnoses were similar between children and adults, with DMD, RYR1, CAPN3, PYGM, DYSF, and FKRP accounting for 50% and Discussion 58% of diagnoses in children and adults, respectively. The previous large LGMD US study identified CAPN3, DYSF, and Gene panels and exome sequencing are increasingly being FKRP as the 3 major contributing genes, associated with 17%, used early in the clinical investigation of patients with sus- 16%, and 9% of diagnoses, respectively, but DMD was only pected muscle disorders, evaluated in outpatient clinics. This found in 4%, whereas RYR1 and PYGM were not tested.7 is likely to be due in part to decreased sequencing cost, im- proved availability of molecular testing, its less invasive nature In our study, variants in DMD were the most frequent cause compared with muscle biopsy, and some previous studies identified on gene panel in both age groups. This relatively reporting high diagnostic yield of such approaches in some high proportion might reflect some differences between our – study populations.26 28 In our study, we report an overall cohort and other published cohorts as regards prior molecular diagnostic yield of 15.1% (22.4% in children and 13.7% in testing to rule out DMD pathogenic variants when there was adults) in a large cohort of patients with suspected muscle a clinical suspicion of Duchenne or Becker dystrophy. We did disorders from outpatient clinics across Canada. Although this not require that DMD deletions/duplications be ruled out by – is lower than most previous studies,2 7 the present estimation specific testing prior performing the gene panel. Despite the is more likely to be representative of current clinical practice, wide availability of multiplex ligation-dependent probe as it relies on less restrictive criteria for testing (such as pre- amplification-based deletion and duplication analysis across vious abnormal muscle biopsy or limb-girdle weakness pat- Canada, a total of 12 of 28 causal variants identified in DMD tern), does not focus on suspicion of a specific group of were deletions of one or multiple exons. We suspect that disorders (e.g., LGMDs), and involves a large number of a change in clinical practice might have occurred among the different physicians (187) from the common medical spe- users of our gene panel as a result of its easy access and its cialties involved in the care of patients with suspected muscle ability to identify virtually all DMD variants in males. This disorders: neuromuscular specialists, general neurologists, would have led to a decrease in the use of MLPA-based de- and geneticists. As expected, the majority of patients (61.7%) letion and duplication analysis and an increase in the pro- were seen by a neuromuscular specialist. The trends sug- portion of DMD cases identified by the 89-gene panel. With gesting some difference in diagnostic yield between those decreasing cost of gene panel and easier availability, this specialists could either reflect referral bias or more stringent tendency could occur in other countries as well. selection of patients for molecular testing. However, all exon deletions were observed in patients with Our diagnostic yield could be underestimated, as some addi- Becker muscular dystrophy and not in patients with Duchenne tional information on patients with potential diagnoses, which muscular dystrophy, which may be related to the more variable

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG phenotype of Becker muscular dystrophy. When detailed Finally, with decreasing sequencing cost, many clinical lab- phenotypic data were available, we could observe suggestive oratories are considering moving to larger sequencers that evidence in some patients that atypical clinical presentation would enable analysis of exome-based virtual panels instead likely contributed to the decision not to request DMD MLPA, of gene panels. Both methods have their advantages and leading to an increased prevalence of deletions observed in our limitations (table e-6, links.lww.com/NXG/A239). In our cohort. For instance, there was a 68-year-old male patient with study, we investigated how this applies in the context of mild limb-girdle and distal weakness and negative family history suspected muscle disorders and the genes that were included (patient 52 on table e-2, links.lww.com/NXG/A235) and two in our analysis. First, there was only minimal difference 11-year-old twins with predominant distal weakness (patients (0.03%) in terms of percentage of bases with adequate 77–78 on table e-2; elevated CK 560–1,100 IU/L, no muscle coverage (>10x), even if median coverage was more than biopsy performed). Another male patient, in whom a DMD 3–10 times higher for gene panel (depending on the se- deletion was subsequently identified, presented initially with quencing platform used) compared with exome-based vir- polyhydramnios and neonatal hypotonia. At age 13 years, he tual panel. In a minority of patients’ samples analyzed by was reported to show diffuse weakness, decreased muscle bulk, exome, some exons showed borderline or low coverage that and negative family history (patient 2 on table e-2; elevated CK could impede detection of single nucleotide variations or 950 IU/L, no biopsy available). In particular, in this last case, indels, although it would not be expected to affect signifi- a variant in a second gene (or an additional nongenetic disor- cantly the diagnostic yield based on causal variants found in der) likely contributed to the atypical phenotype. the gene panel cohort. Moreover, detection of large het- erozygous deletions and duplications could be limited by Finally, the fact that DMD sequencing is not easily accessible lower coverage of the exome-based panel, especially when in some Canadian provinces and requires special approval dropping below 50x.24 Nonetheless, even with optimal processes that may limit access in some cases may have coverage, detection of small exon deletions (<3 exons) contributed to an increased proportion of DMD cases in our remains challenging.35,36 This applies to DMD heterozygous cohort. Supporting this, all patients diagnosed with Duchenne deletions in female and duplications in both male and fe- muscular dystrophy were caused by single nucleotide varia- male. In our study, we did not systematically study the tion. Those patients are expected to be recognized on a clin- performance of both MLPA and next-generation sequencing ical basis, and thus would likely not be sent for gene panel if method to detect those variants. As deletions and duplica- DMD sequencing was more accessible in some Canadian tions remain a rare type of causal variant, apart from the provinces. This factor presumably contributed to increase the X-linked DMD gene exon deletions that can still be detected proportion of causal DMD variants observed, especially in the with decreasing coverage in males, the impact on the di- pediatric group compared with other countries having easier agnostic yield would have been limited. Still, it would have access to DMD sequencing. Nevertheless, even if we exclude possibly resulted in 1 missed diagnosis (1/187, 0.5%). This patients with Duchenne muscular dystrophy, variants in DMD emphasizes the need to carefully validate each virtual gene would remain the most common cause of muscle disorders panel to ensure sufficient coverage to detect all variant types identified in the pediatric group. with optimal performance. Alternatively, performing addi- tional deletion analysis in parallel by exon array or MLPA for We investigated whether any clinical criteria could be used to specific genes could be recommended when a single path- select patients who would more likely benefit from gene panel ogenic variant is found or for DMD in all cases. Nevertheless, testing while having minimal risk of missing a diagnosis. As despite these limitations, our study shows that an exome- expected, positive family history and younger age were cor- based panel with analysis extended to other mendelian genes related with an increased likelihood of finding a genetic eti- increased the diagnostic yield from 15.2% to 21.7% in the ology, but the yield remained significant in patients older than exome cohort. On patients selected with the same criteria, 60 years (9.9%). We had to use age at testing as a proxy for age the 3 additional diagnoses emphasize the fact that all gene at onset because accurate historical information was not sys- panel designs are subject to omission of genes that may tematically available. Other criteria that were associated with represent more atypical cause of weakness (HADHA), the increased diagnostic yield were high serum CK (especially if difficulty sometimes encountered in differentiating myo- above 1,000 IU/L), the presence of a cardiac anomaly, and pathic from neuropathic pathologies on a clinical basis a positive familial history. These likely reflect better diagnostic (MYH14), and accuracy for triplet repeats might be limited yield in muscular dystrophies.31,33,34 Nevertheless, excluding for next-generation sequencing compared with PCR patients with normal CK results and no cardiac anomaly from (PABPN1).37 In particular, our panel did not provide com- testing would have resulted in missing 10.7% of the diagnoses, prehensive coverage of metabolic myopathies, given that the including 3/6 patients with MYOT causal variants who were primary indication of our panel was weakness and not aged >55 years. Some other patients with such conditions may rhabdomyolysis. It is not possible to transpose and gener- have been excluded from testing by our protocol, which re- alize this increase of diagnostic yield to the group of patients quired 1 abnormal laboratory for testing to be performed. recruited from various Canadian outpatient clinics (cohort Evaluation of more patients would be needed to analyze the of 1,236 patients), although overall diagnostic yield of the 89 impacts of more complex combinations of clinical criteria. genes is similar between the 2 groups of patients studied

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 (15.1% and 15.2%). Various bias could influence the pro- specialists, genetics counsellor, and clinical geneticists across portion of cases harbouring variants in genes outside the 89- Canada for their participation in this study. F. Thuriot is gene panel. This includes a different population of recruiting supported by doctoral scholarship from the Fonds de physicians with different level of expertise in muscle dis- recherche du Qu´ebec–Sant´e (FRQS). orders and a different proportion of abnormal muscle bi- opsies or prior negative genetic tests, which are both not Study funding accurately known in the cohort of 1,236 patients. We also This study was sponsored by Sanofi Genzyme Canada. observed a modest increase of variants of uncertain clinical significance related to patient phenotype when extending Disclosure the analysis beyond the virtual 89-gene panel. However, we Disclosures available: Neurology.org/NG. were quite conservative in reporting and had access to the complete patient file in the exome cohort. The number of Publication history VUS might be much more increased in the regular laboratory Received by Neurology: Genetics May 31, 2019. Accepted in final form setting with limited clinical information. Overall, our results January 24, 2020. support the use of an exome-based panel for muscular dis- orders, although several considerations must be taken into account when introducing exome-based panel in the – Appendix Authors laboratory38 40 (table e-6, links.lww.com/NXG/A239). Name Location Contribution

Establishing a molecular diagnosis in patients with muscle dis- Fanny Universit´ede Analyzed the data and drafted the orders is becoming increasingly important, as the potential to Thuriot, MSc Sherbrooke, manuscript for intellectual Quebec, Canada content alterthediseasecoursewithgenespecific treatments is expected to increase in the upcoming years. Specific treatments are cur- Elaine Universit´ede Analyzed the data Gravel, MSc Sherbrooke, rently limited for hereditary muscle disorders, and enzyme re- Quebec, Canada placement therapy represents 1 example. However, gene Caroline Universit´ede Analyzed the data therapy approaches are undergoing rapid development for the Buote, MSc Sherbrooke, treatment of LGMDs and dystrophinopathies, in particular. Quebec, Canada

Indeed, several studies are in clinical trials and have shown the Marianne Universit´ede Acquisition of data 41–46 potential of this method. This shows the importance of Doyon, MD Sherbrooke, having a diagnosis for these patients. Quebec, Canada Elvy Universit´ede Acquisition of data Our study supports the use of gene panel testing in patients Lapointe, Sherbrooke, MSc Quebec, Canada with suspected muscle disorders from outpatient clinics and highlights several relatively common diagnoses identified in Lydia Universit´ede Acquisition of data Marcoux, Sherbrooke, adults and children (DMD, RYR1, CAPN3, PYGM, DYSF, and MSc Quebec, Canada FKRP). Moreover, it shows that exome sequencing has a low Sandrine Universit´ede Acquisition of data risk of missing diagnoses compared with gene panel, while Larue, MD Montr´eal, Quebec, potentially increasing the diagnostic yield of patients with Canada

muscle disorders. However, exome sequencing comes with Am´elie Universit´ede Acquisition of data a bigger burden of VUS. It would though be important, with an Nadeau, MD Sherbrooke, exome approach, to have elaborated clinical data and to have Quebec, Canada the data analyzed in a center experimented in variant in- S´ebastien Universit´ede Revised the manuscript for terpretation in neuromuscular conditions. Combining MLPA- Ch´enier, MD Sherbrooke, intellectual content Quebec, Canada based deletion and duplication analysis for DMD with 1 of the 2 sequencing methods could result in a higher accuracy for Du- Paula J. Universit´ede Revised the manuscript for Waters, PhD Sherbrooke, intellectual content chenne and Becker muscular dystrophies, especially in females, Quebec, Canada given lower sensibility for small heterozygous exon deletions. Pierre- Universit´ede Analyzed the data ´Etienne Sherbrooke, Acknowledgment Jacques, PhD Quebec, Canada fi The study was supported by Sano Genzyme Canada. The Serge Gravel, Universit´ede Designed and conceptualized the authors are thankful to G´enome Qu´ebec and Fulgent for the PhD Sherbrooke, study; analyzed the data; and Quebec, Canada revised the manuscript for exome sequencing, Dynacare for their contribution in intellectual content samples’ logistics and genetic counselling, Jean-François Lussier for his computing support, and Calcul Qu´ebec and S´ebastien Universit´ede Designed and conceptualized the L´evesque, Sherbrooke, study; analyzed the data; and drafted Compute Canada for providing part of the computing MD, PhD Quebec, Canada the manuscript infrastructure used to analyze the data. They are also grateful for intellectual content to the patients and their families, neurologists, neuromuscular

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG References 25. Richards S, Aziz N, Bale S, et al. Standards and guidelines for the interpretation of 1. Ghaoui R, Clarke N, Hollingworth P, Needham M. Muscle disorders: the latest sequence variants: a joint consensus recommendation of the American College of investigations. Intern Med J 2013;43:970–978. doi:10.1111/imj.12234. Medical Genetics and Genomics and the Association for Molecular Pathology. Genet – 2. Yu M, Zheng Y, Jin S, et al. Mutational spectrum of Chinese LGMD patients by Med 2015;17:405 423. doi:10.1038/gim.2015.30. targeted next-generation sequencing. PLoS One 2017;12:e0175343. doi:10.1371/ 26. Volk AE, Kubisch C. The rapid evolution of molecular genetic diagnostics in neu- – journal.pone.0175343. romuscular diseases. Curr Opin Neurol 2017;30:523 528. doi:10.1097/WCO. 0000000000000478. 3. Kuhn M, Gl¨aser D, Joshi PR, et al. Utility of a next-generation sequencing-based fi ff gene panel investigation in German patients with genetically unclassified limb- 27. Scho eld D, Alam K, Douglas L, et al. Cost-e ectiveness of massively parallel se- quencing for diagnosis of paediatric muscle diseases. NPJ Genomic Med 2017;2. doi: girdle muscular dystrophy. J Neurol 2016;263:743–750. doi:10.1007/s00415- 016-8036-0. 10.1038/s41525-017-0006-7. 28. Ankala A, da Silva C, Gualandi F, et al. A comprehensive genomic approach for 4. Seong M-W, Cho A, Park HW, et al. Clinical applications of next-generation neuromuscular diseases gives a high diagnostic yield. Ann Neurol 2015;77:206–214. sequencing-based gene panel in patients with muscular dystrophy: Korean experience. doi:10.1002/ana.24303. Clin Genet 2016;88:484–488. doi:10.1111/cge.12621. 29. Stehl´ıkov´a K, Sk´alov´aD,Z´ıdkov´a J, et al. Muscular dystrophies and myopathies: the 5. Kitamura Y, Kondo E, Urano M, Aoki R, Saito K. Target resequencing of neuro- spectrum of mutated genes in the Czech Republic. Clin Genet 2017;91:463–469. doi: muscular disease-related genes using next-generation sequencing for patients with 10.1111/cge.12839. undiagnosed early-onset neuromuscular disorders. J Hum Genet 2016;61:931–942. 30. Vasli N, Laporte J. Impacts of massively parallel sequencing for genetic diagnosis of doi:10.1038/jhg.2016.79. neuromuscular disorders. Acta Neuropathol (Berl) 2013;125:173–185. doi:10.1007/ 6. L´evesque S, Auray-Blais C, Gravel E, et al. Diagnosis of late-onset Pompe disease and s00401-012-1072-7. other muscle disorders by next-generation sequencing. Orphanet J Rare Dis 2016;11: 31. Magri F, Nigro V, Angelini C, et al. The Italian limb girdle muscular dystrophy 8. doi:10.1186/s13023-016-0390-6. registry: relative frequency, clinical features, and differential diagnosis. Muscle Nerve 7. Nallamilli BRR, Chakravorty S, Kesari A, et al. Genetic landscape and novel disease 2017;55:55–68. doi:10.1002/mus.25192. mechanisms from a large LGMD cohort of 4656 patients. Ann Clin Transl Neurol 32. Park HJ, Jang H, Kim JH, et al. Discovery of pathogenic variants in a large Korean cohort – 2018;5:1574 1587. doi:10.1002/acn3.649. of inherited muscular disorders. Clin Genet 2017;91:403–410. doi:10.1111/cge.12826. 8. Thuriot F, Buote C, Gravel E, et al. Clinical validity of phenotype-driven analysis 33. Te Riele MGE, Schreuder THA, van Alfen N, et al. The yield of diagnostic work-up of software PhenoVar as a diagnostic aid for clinical geneticists in the interpretation of patients presenting with myalgia, exercise intolerance, or fatigue: a prospective obser- – whole-exome sequencing data. Genet Med 2018;20:942 949. doi:10.1038/gim.2017. vational study. Neuromuscul Disord 2017;27:243–250. doi:10.1016/j.nmd.2016.12.002. 239. 34. Narayanaswami P. Evidence-based guideline summary: Diagnosis and treatment of 9. Haskell GT, Adams MC, Fan Z, et al. Diagnostic utility of exome sequencing in the limb-girdle and distal dystrophies: Report of the Guideline Development Sub- evaluation of neuromuscular disorders. Neurol Genet 2018;4:e212. doi:10.1212/ committee of the American Academy of Neurology and the Practice Issues Review NXG.0000000000000212. Panel of the American Association of Neuromuscular & Electrodiagnostic Medicine. 10. Retterer K, Juusola J, Cho MT, et al. Clinical application of whole-exome sequencing Neurology 2014;83:1453–1463. – across clinical indications. Genet Med 2016;18:696 704. doi:10.1038/gim.2015.148. 35. Wang J, Yu H, Zhang VW, et al. Capture-based high-coverage NGS: a powerful tool to 11. Harris E, Topf A, Barresi R, et al. Exome sequences versus sequential gene testing in uncover a wide spectrum of mutation types. Genet Med 2016;18:513–521. doi:10. the UK highly specialised Service for limb girdle muscular dystrophy. Orphanet J Rare 1038/gim.2015.121. Dis 2017;12:151. doi:10.1186/s13023-017-0699-9. 36. Yao R, Yu T, Qing Y, Wang J, Shen Y. Evaluation of copy number variant detection 12. Reddy HM, Cho K-A, Lek M, et al. The sensitivity of exome sequencing in identifying from panel-based next-generation sequencing data. Mol Genet Genomic Med 2019;7: pathogenic mutations for LGMD in the United States. J Hum Genet 2017;62: e00513. doi:10.1002/mgg3.513. 243–252. doi:10.1038/jhg.2016.116. 37. Keogh MJ, Chinnery PF. Next generation sequencing for neurological diseases: new 13. Ghaoui R, Cooper ST, Lek M, et al. Use of whole-exome sequencing for diagnosis of hope or new hype? Clin Neurol Neurosurg 2013;115:948–953. doi:10.1016/j.cli- limb-girdle muscular dystrophy: outcomes and lessons learned. JAMA Neurol 2015; neuro.2012.09.030. 72:1424–1432. doi:10.1001/jamaneurol.2015.2274. 38. Bertier G, H´etu M, Joly Y. Unsolved challenges of clinical whole-exome sequencing: 14. Fichna JP, Macias A, Piechota M, et al. Whole-exome sequencing identifies novel a systematic literature review of end-users’ views. BMC Med Genomics 2016;9:52. pathogenic mutations and putative phenotype-influencing variants in Polish limb- doi:10.1186/s12920-016-0213-6. girdle muscular dystrophy patients. Hum Genomics 2018;12:34. doi:10.1186/ 39. Sie AS, Prins JB, van Zelst-Stams WAG, Veltman JA, Feenstra I, Hoogerbrugge N. s40246-018-0167-1. Patient experiences with gene panels based on exome sequencing in clinical diag- 15. Reddy HM, Hamed SA, Lek M, et al. Homozygous nonsense mutation in SGCA is nostics: high acceptance and low distress. Clin Genet 2015;87:319–326. doi:10.1111/ a common cause of limb-girdle muscular dystrophy in Assiut, Egypt: LGMD in Egypt cge.12433. Linked to SGCA. Muscle Nerve 2016;54:690–695. doi:10.1002/mus.25094. 40. Wang X, Shen X, Fang F, et al. Phenotype-driven virtual panel is an effective method 16. Efthymiou S, Manole A, Houlden H. Next-generation sequencing in neuromuscular to analyze WES data of neurological disease. Front Pharmacol 2018;9:1529. doi:10. diseases. Curr Opin Neurol 2016;29:527–536. doi:10.1097/WCO.0000000000000374. 3388/fphar.2018.01529.8888. 17. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence 41. Mendell JR, Chicoine LG, Al-Zaidy SA, et al. Gene delivery for limb-girdle muscular data. Bioinformatics 2014;30:2114–2120. doi:10.1093/bioinformatics/btu170. dystrophy type 2D by isolated limb infusion. Hum Gene Ther 2019;30:794–801. doi: 18. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler 10.1088/hum.2019.006. transform. Bioinformatics 2009;25:1754–1760. doi:10.1093/bioinformatics/btp324. 42. Nelson CE, Robinson-Hamm JN, Gersbach CA. Genome engineering: a new ap- 19. DePristo MA, Banks E, Poplin R, et al. A framework for variation discovery and proach to gene therapy for neuromuscular disorders. Nat Rev Neurol 2017;13: genotyping using next-generation DNA sequencing data. Nat Genet 2011;43: 647–661. doi:10.1038/nrneurol.2017.126. 491–498. doi:10.1038/ng.806. 43. Vannoy CH, Xu L, Keramaris E, Lu P, Xiao X, Lu QL. Adeno-associated virus- 20. Cingolani P, Platts A, Wang LL, et al. A program for annotating and predicting the mediated overexpression of LARGE rescues α-dystroglycan function in dystrophic effects of single nucleotide polymorphisms, SnpEff: SNPs in the genome of Dro- mice with mutations in the fukutin-related protein. Hum Gene Ther Methods 2014; 1118 sophila melanogaster strain w ; iso-2; iso-3. Fly (Austin) 2012;6:80–92. doi:10. 25:187–196. doi:10.1088/hgtb.2013.151. 4161/fly.19695. 44. Sondergaard PC, Griffin DA, Pozsgai ER, et al. AAV.Dysferlin overlap vectors restore 21. Landrum MJ, Lee JM, Benson M, et al. ClinVar: public archive of interpretations of clinically function in dysferlinopathy animal models. Ann Clin Transl Neurol 2015;2:256–270. relevant variants. Nucleic Acids Res 2016;44:D862–D868. doi:10.1093/nar/gkv1222. doi:10.1002/acn3.172. 22. Lek M, Karczewski KJ, Minikel EV, et al. Analysis of protein-coding genetic variation 45. Bengtsson NE, Hall JK, Odom GL, et al. Muscle-specific CRISPR/Cas9 dystrophin in 60,706 humans. Nature 2016;536:285–291. doi:10.1038/nature19057. gene editing ameliorates pathophysiology in a mouse model for Duchenne muscular 23. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing genomic features. dystrophy. Nat Commun 2017;8:14454. doi:10.1038/ncomms14454. Bioinformatics 2010;26:841–842. doi:10.1093/bioinformatics/btq033. 46. Nelson CE, Hakim CH, Ousterout DG, et al. In vivo genome editing improves muscle 24. Johansson LF, van Dijk F, de Boer EN, et al. CoNVaDING: single exon variation detection function in a mouse model of Duchenne muscular dystrophy. Science 2016;351: in targeted NGS data. Hum Mutat 2016;37:457–464. doi:10.1002/humu.22969. 403–407. doi:10.1126/science.aad5143.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 ARTICLE OPEN ACCESS Hereditary cerebral amyloid angiopathy, Piedmont-type mutation

Mariel G. Kozberg, MD, PhD, Susanne J. van Veluw, PhD, Matthew P. Frosch, MD, PhD, and Correspondence Steven M. Greenberg, MD, PhD Dr. Kozberg [email protected] Neurol Genet 2020;6:e411. doi:10.1212/NXG.0000000000000411 Abstract Objective We present here a case report of a patient with a family history of intracerebral hemorrhages (ICHs) who presented with multiple large lobar hemorrhages in rapid succession, with cog- nitive sparing, who was found to have a mutation in the β-amyloid coding sequence of amyloid precursor protein (Leu705Val), termed the Piedmont-type mutation, the second ever reported case of this form of hereditary cerebral amyloid angiopathy (CAA).

Methods Targeted pathologic examination was performed aided by the use of ex vivo MRI.

Results Severe CAA was observed mainly involving the leptomeningeal vessels and, to a far lesser extent, cortical vessels, with no amyloid plaques or neurofibrillary tangles.

Conclusions This leptomeningeal pattern of β-amyloid deposition coupled with multiple large hemorrhages demonstrates unique pathophysiologic characteristics of CAA associated with the Piedmont- type mutation, suggesting a potential association between leptomeningeal CAA and larger ICHs.

From the MassGeneral Institute for Neurodegenerative Disease (M.G.K., S.J.v.V.), Massachusetts General Hospital and Harvard Medical School, Charlestown; Department of Neurology (M.G.K., S.J.v.V., S.M.G.), Massachusetts General Hospital, Boston; Department of Neurology (M.G.K.), Brigham and Women’s Hospital, Boston; J. Philip Kistler Stroke Research Center (S.J.v.V., S.M.G.), Massachusetts General Hospital and Harvard Medical School, Boston; and Neuropathology Service, C. S. Kubik Laboratory for Neuropathology (M.P.F), Massachusetts General Hospital and Harvard Medical School, Boston.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary APP = amyloid precursor protein; CAA = cerebral amyloid angiopathy; GRE = gradient echo; ICH = intracerebral hemorrhage; TSE = turbo spin echo.

Cerebral amyloid angiopathy (CAA) is a disorder in which without neurofibrillary tangles or β-amyloid plaques within the β-amyloid peptide is deposited in leptomeningeal and cortical brain parenchyma. vessels of the brain. This deposition is associated with both cortical microbleeds and large lobar intracerebral hemorrhages We present here a second, independent case with similar (ICHs); however, precise mechanisms leading to vessel rupture pathologic observations and focus on unique clinical and remain undetermined. Although some degree of CAA is pathologic features not previously described. commonly observed at routine neuropathologic examination, hereditary CAA is relatively rare and typically more severe than Methods sporadic CAA. A number of mutations involving both the coding and noncoding regions of amyloid precursor protein Neuropathologic examination and ex vivo MRI (APP) have been described, each with distinct phenotypes and Brain autopsy was performed after organ donation (with – pathologic features.1 4 a 67-hour postmortem interval before brain extraction). The brain was fixed in 10% formalin, and the hemispheres were The Piedmont-type mutation is a mutation in the coding region separated by a midsagittal cut. The left hemisphere un- of APP (Leu705Val).5 The single reported postmortem brain derwent standard neuropathologic evaluation. The right with this mutation demonstrated vascular β-amyloid pathology hemisphere was fixed for 2 months in 10% formalin and then

Figure 1 Clinical data

(A) CT scans demonstrating the time course of patient’s multiple lobar ICHs. (B) Pedigree of proband’s family. Black circles/squares represent family members with ICHs. Gray square represents a family member who died of stroke in his 60s, unknown if stroke was hemorrhagic or ischemic. Diamond symbol indicates family members with autopsies confirming cerebral amyloid angiopathy. *Age at time of first symptomatic hemorrhage. ICH = intracerebral hemorrhage.

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 2 Pathologic examination

(A) β-amyloid staining revealed severe widespread leptomeningeal CAA with relative sparing of the cortical vessels. CAA was observed in capillaries in the occipital lobe (bottom right panel). (B) Cortical microinfarct and cortical microbleed shown with both H&E and β-amyloid staining. β-amyloid staining demonstrated β-amyloid deposition in the walls of the vessels supplying the infarcted area (denoted by arrows). No β-amyloid staining was observed in the segment of vessel with a microbleed (denoted by arrow). CAA = cerebral amyloid angiopathy; H&E = hematoxylin and eosin. scanned intact with high-resolution ex vivo 3T MRI, using Data availability gradient echo (GRE) and turbo spin echo (TSE) sequences, All data not published within this article will be shared on as previously described.6 request from any qualified investigator.

Afterward, the right hemisphere was cut in coronal slabs and the samples were taken from frontal, parietal, temporal, and occipital Results lobes for histopathologic evaluation including standard hematox- Clinical data ylin and eosin (H&E) staining and immunohistochemistry against A 62-year-old right-handed woman with a history of paroxysmal β-amyloid (clone 6F/3D, Agilent). Additional samples were taken atrial fibrillation (not on anticoagulation or antiplatelet agents) from the areas with cortical microbleeds on MRI, and serial sec- presented with a severe right-sided retro-orbital headache, tioning was performed to evaluate the ruptured blood vessels. word-finding difficulty, and issues with balance. She reported – Genetic analysis one previous episode of 1 2hoursofconfusioninthepreceding Genetic sequencing was performed through Athena Diag- months and 2 previous episodes of spreading sensory symptoms nostics ADmark APP DNA Sequencing Test and Duplication involving tingling on her left side in varied distributions, each Test (test code 168), which confirmed a heterozygous mis- lasting approximately 10 minutes. A CT head demonstrated fi sense mutation in APP (c.2113C>G; p.Leu705Val). a right-sided parietal ICH ( gure 1A, upper left panel) and a remote right frontal ischemic infarct. After this event, her Standard protocol approvals, registrations, symptoms fully resolved and she continued working as a nurse. and patient consents The Massachusetts General Hospital institutional review board Over the next 3 months, she was hospitalized 4 more times for approved this study. Autopsy was performed with informed recurrent lobar ICHs (right temporal, left cingulate/left consent of the patient’sfamily. frontal, right medial temporal, and left frontal respectively; figure 1A). After each hospitalization, her cognition was ini- Family history was obtained through interviews with family tially impaired but quickly improved to close to her baseline. members. Four months after initial presentation, she was found down and

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 a CT head demonstrated a new right temporoparietal ICH with tau using AT8 antibody was negative. No evidence of in- associated midline shift (figure 1A; bottom right panel). Her flammation was observed. No hypertensive arteriopathy was examination was consistent with brain death. observed in the larger vessels or the basal ganglia.

The proband had a family history of ICH consistent with The GRE scan of the right hemisphere revealed 51 cortical autosomal dominant inheritance, including her grandfather, microbleeds and 3 sites of cortical superficial siderosis. These her father and 2 of his siblings, and 2 of the patient’s 3 siblings hemorrhagic lesions were distributed across the cortex, with (figure 1B). Her paternal family ancestry traces back to no specific lobar predilection (figure 3). The detection of England and Wales, with no known Italian lineage or overlap cortical microinfarcts on TSE was hampered by widespread with the single previously reported family with the Piedmont- diffuse hypoxic-ischemic tissue injury related to the long type mutation5; however, shared ancestry was not assessed postmortem interval in the setting of organ donation. through further genetic analysis. Similar to the proband, all affected family members were working and at their cognitive H&E-stained sections from each lobe were screened for old baselines before their initial ICHs. Autopsy results of the microinfacts. Four cortical microinfarcts were identified in total proband’s aunt confirmed CAA. Genetic testing has not been (across all 4 sections), fewer than typically observed in cases with performed on other family members to our knowledge. sporadic CAA using similar methodology.8 The cortical vessels associated with each microinfarct had substantial β-amyloid in their Pathologic examination walls (figure 2B). Serial sectioning of MRI-guided samples yielded The brain weight on autopsy was within the normal range for 3 microbleeds (2 recent, one old). The walls of vessels associated women (1,550 g), with no evidence of significant atrophy. with microbleeds did not contain β-amyloid (figure 2B). Immunohistochemistry against β-amyloid revealed widespread severe CAA, most pronounced in leptomeningeal vessels, al- though also observed in cortical vessels and capillaries (figure Discussion 2A). Vascular thickening and focal splitting of the vessel wall, consistent with Vonsattel grade 3/4,7 were frequently observed The neuropathologic evidence shown here, together with in the leptomeningeal vessels and occasionally in the cortical a previously reported unrelated individual,5 suggests that the vessels. Immunohistochemistry against hyperphosphorylated Piedmont-type mutation in APP leads to a distinct form of

Figure 3 High-resolution ex vivo MRI

Top, ex vivo GRE scan, sagittal view. Large arrows point to lobar ICHs, smaller arrows to cortical microbleeds. Bottom, 3D representation of cortical microbleeds (red) and larger ICH (yellow) locations, insets with scans of the cortical surface. GRE = gradient echo; ICH = intracerebral hemorrhage.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG hereditary CAA with exclusively vascular β-amyloid de- provide valuable insights into the mechanisms of β-amyloid position. We present a patient with a confirmed Leu705Val deposition and vessel damage. mutation in APP, who suffered from multiple large lobar ICHs in rapid succession, with no preceding cognitive decline and Acknowledgment good cognitive recoveries from her hemorrhages until her The authors thank Karen Bechdol for researching and sharing terminal event. Affected family members had similar clinical the family history presented in this manuscript. courses, with no cognitive decline preceding their lobar ICHs. This clinical course is similar to that of the previously reported Study funding Italian family with this mutation5 and distinct from other The authors acknowledge support from the National Insti- hereditary forms of CAA, suggesting that it is characteristic of tutes of Health grants: R01AG26484 (Greenberg), this mutation. P30AG062421 (Frosch), and K99AG059893 (van Veluw).

We demonstrated that β-amyloid deposits were primarily Disclosure found within the leptomeningeal vessels, with relative sparing Disclosures available: Neurology.org/NG. of the cortical vasculature. Based on this finding and the rapid pace of recurrent hemorrhages, we hypothesize that extensive Publication history leptomeningeal involvement may predispose patients with the Received by Neurology: Genetics November 27, 2019. Accepted in final Piedmont-type mutation to a more severe hemorrhagic phe- form January 27, 2020. notype than sporadic CAA and some other hereditary forms of CAA. Appendix Authors

The MRI-guided neuropathologic analysis offers potential Name Location Contribution insights into the mechanisms of CAA-related hemorrhage. We Mariel Brigham and Women’s Data collection and examined vessels associated with cortical microbleeds and Kozberg, Hospital; Massachusetts analysis, drafting, and observed no β-amyloid in the vessels at the bleeding site. This MD, PhD General Hospital; Harvard revision of the manuscript. finding is consistent with previous observations in sporadic Medical School, Boston 8,9 CAA, suggesting a mechanism in which severely affected Susanne Massachusetts General Data collection and β van Veluw, Hospital; Harvard Medical analysis, drafting, and vessel segments undergo remodeling entailing loss of -am- PhD School, Boston revision of the manuscript. yloid deposits before hemorrhage.8 Determining the exact Matthew Massachusetts General Data collection and rupture site for leptomeningeal bleeds is more challenging Frosch, MD, Hospital; Harvard Medical analysis, revision of the because these hemorrhages are often larger and more dis- PhD School, Boston manuscript. persed with significant damage to the surrounding tissue. It is Steven Massachusetts General Clinical care, data therefore unclear whether leptomeningeal vessel hemor- Greenberg, Hospital; Harvard Medical collection and analysis, rhages follow the same proposed pathophysiologic mecha- MD, PhD School, Boston drafting, and the revision of manuscript. nism as smaller cortical microbleeds, a topic for future investigations. References 1. Biffi A, Greenberg SM. Cerebral amyloid angiopathy: a systematic review. J Clin Qualitative assessment of H&E-stained sections suggested – 8 Neurol 2011;7:1 9. fewer chronic microinfarcts than typical in sporadic CAA. 2. Bornebroek M, Jonghe CDe, Haan J, et al. Hereditary cerebral hemorrhage with Similar to observations in sporadic CAA, the vessels feeding amyloidosis Dutch type (AbetaPP 693): decreased plasma amyloid-beta42 concen- tration. Neurobiol Dis 2003;14:619–623. the infarcted tissue demonstrated increased local β-amyloid 3. Greenberg SM, Shin Y, Grabowski TJ, et al. Hemorrhagic stroke associated with the burden.8 These findings suggest that the relative sparing of Iowa amyloid precursor protein mutation. Neurology 2003;60:1020–1022. 4. De Jonghe C, Zehr C, Yager D, et al. Flemish and Dutch mutations in amyloid beta the cortical vasculature from CAA may lead to fewer precursor protein have different effects on amyloid beta secretion. Neurobiol Dis microinfarcts in patients with the Piedmont mutation, 1998;5:281–286. 5. Obici L, Demarchi A, De Rosa G, et al. A novel AβPP mutation exclusively associated a factor likely contributing to the lack of cognitive impair- with cerebral amyloid angiopathy. Ann Neurol 2005;58:639–644. ment in these patients (in addition to the absence of plaques 6. van Veluw SJ, Reijmer YD, Van Der Kouwe AJ, et al. Histopathology of diffusion imaging abnormalities in cerebral amyloid angiopathy. Neurology 2019;92:E933–E943. and tangles). 7. Vonsattel JPG, Myers RH, Tessa Hedley‐Whyte E, Ropper AH, Bird ED, Richardson EP. Cerebral amyloid angiopathy without and with cerebral hemorrhages: a compar- fi ative histological study. Ann Neurol 1991;30:637–649. Our ndings demonstrate that the Piedmont-type mutation 8. van Veluw SJ, Scherlek AA, Freeze WM, et al. Different microvascular alterations leads to specific accumulation of β-amyloid within lep- underlie microbleeds and microinfarcts. Ann Neurol 2019;86:279–292. 9. van Veluw SJ, Kuijf HJ, Charidimou A, et al. Reduced vascular amyloid burden at tomeningeal vessels. Further investigation into the molecular microhemorrhage sites in cerebral amyloid angiopathy. Acta Neuropathol 2017;133: properties of the mutated β-amyloid has the potential to 409–415.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 ARTICLE OPEN ACCESS Clinical utility of multigene analysis in over 25,000 patients with neuromuscular disorders

Thomas L. Winder, PhD, Christopher A. Tan, MS, Sarah Klemm, MS, Hannah White, MS, Correspondence Jody M. Westbrook, PhD, James Z. Wang, PhD, Ali Entezam, PhD, Rebecca Truty, PhD, Robert L. Nussbaum, MD, Dr. Winder [email protected] Elizabeth M. McNally, MD, PhD, and Swaroop Aradhya, PhD

Neurol Genet 2020;6:e412. doi:10.1212/NXG.0000000000000412 Abstract Objective Molecular genetic testing for hereditary neuromuscular disorders is increasingly used to identify disease subtypes, determine prevalence, and inform management and prognosis, and although many small disease-specific studies have demonstrated the utility of genetic testing, compre- hensive data sets are better positioned to assess the complexity of genetic analysis.

Methods Using high depth-of-coverage next-generation sequencing (NGS) with simultaneous detection of sequence variants and copy number variants (CNVs), we tested 25,356 unrelated individuals for subsets of 266 genes.

Results Adefinitive molecular diagnosis was obtained in 20% of this cohort, with yields ranging from 4% among individuals with congenital myasthenic syndrome to 33% among those with a muscular dystrophy. CNVs accounted for as much as 39% of all clinically significant variants, with 10% of them occurring as rare, private pathogenic variants. Multigene testing successfully addressed differential diagnoses in at least 6% of individuals with positive results. Even for classic disorders like Duchenne muscular dystrophy, at least 49% of clinically significant results were identified through gene panels intended for differential diagnoses rather than through single-gene analysis. Variants of uncertain significance (VUS) were observed in 53% of indi- viduals. Only 0.7% of these variants were later reclassified as clinically significant, most com- monly in RYR1, GDAP1, SPAST, and MFN2, providing insight into the types of evidence that support VUS resolution and informing expectations of reclassification rates.

Conclusions These data provide guidance for clinicians using genetic testing to diagnose neuromuscular disorders and represent one of the largest studies demonstrating the utility of NGS-based testing for these disorders.

From the Invitae Corporation (T.L.W., C.A.T., S.K., H.W., J.M.W., J.Z.W., A.E., R.T., R.L.N., S.A.), San Francisco, CA; Volunteer Faculty (R.L.N.), University of California, San Francisco; and Center for Genetic Medicine (E.M.M.), Northwestern University, Evanston, IL.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary AD = autosomal dominant; AR = autosomal recessive; BMD = Becker muscular dystrophy; CMT = Charcot-Marie-Tooth; CMT1A = Charcot-Marie-Tooth disease type 1A; CNV = copy number variant; DMD = Duchenne muscular dystrophy; HNPP = hereditary neuropathy with pressure palsies; LB/B = likely benign and benign; LP/P = likely pathogenic or pathogenic; NGS = next-generation sequencing; SMA = spinal muscular atrophy; SNV = single-nucleotide variant; VUS = variants of uncertain significance; XL = X-linked.

Molecular genetic analyses can provide diagnostic clarity for regions are detected. Validation of bioinformatics methods neuromuscular disorders, which demonstrate considerable and the clinical utility of CNV detection in NGS panels have – clinical and genetic heterogeneity.1 4 In some cases, geneti- been previously described.15,16 The analysis of SMN1 and cally disparate neuromuscular disorders share overlapping SMN2 was conducted using a validated bioinformatics algo- pathogenic mechanisms that correlate with and explain the rithm that gathered sequence reads from both gene copies in – subtleties of the clinical presentations.5 8 Next-generation a single bin and subsequently used the presence or absence of sequencing (NGS), enabling simultaneous analysis of many the c.840C>T variant in exon 7 to disambiguate sequence genes without significant additional cost has now made ge- reads and accurately determine copy number at each locus. netic testing far more accessible. Studies using NGS in The algorithm does not determine the phase of SMN1 gene disease-specific contexts have reported high diagnostic yields copies, and thus, silent spinal muscular atrophy (SMA) car- ranging from 19% for spastic paraplegia to 60% for neuro- riers cannot be detected. – muscular disorders.9 11 Historically, copy number variant (CNV) analysis has been limited to only a narrow group of Subjects and reporting criteria genes12 or has not been routinely performed in neuromus- Between October 2014 and April 2019, an unbiased cohort of cular genetic testing.9,13,14 Recently, the accurate detection of patients suspected to have a neuromuscular disorder was re- CNVs alongside sequence variants by NGS in a single assay ferred with informed consent for genetic testing (table 1). Only has enabled more comprehensive, more affordable, and faster index patients were counted in this study. Clinicians requested genetic analysis with a single sample.15 testing for all genes on a panel or chose subpanels for narrower clinical indications. NGS testing and clinical variant in- 17 The purpose of this study was to examine a large unselected terpretation was performed as described previously. Reports clinical cohort with neuromuscular disorders to evaluate the included variants classified as likely pathogenic or pathogenic diagnostic yield of gene panels with simultaneous sequence (LP/P) or VUS; likely benign and benign (LB/B) variants were and CNV detection. In one of the largest studies of genetic not reported. A definitive molecular diagnosis included a single testing for this group of disorders, we also investigated the LP/P variant in a gene associated with autosomal dominant mutation spectrum, mutation properties, and reclassification (AD) or X-linked (XL) inheritance, or 2 variants either in the of variants of uncertain significance (VUS) to obtain insight homozygous or compound heterozygous state in the appro- into genetic aspects of neuromuscular disorders that should priate phase in genes associated with autosomal recessive (AR) inform clinical diagnosis of affected individuals. inheritance. LP/P variants were confirmed using the Sanger or PacBio sequencing for sequence variants and exon-focused microarray or NGS-based multiplex ligation-dependent ampli- Methods fication was performed on CNVs larger than 250–500 bp. All variants for this study were collected from Invitae’sinternal Gene panel design database and annotated based on the guidelines from the Hu- fi Phenotype-speci c gene panels were curated by evaluating man Genome Variation Society (varnomen.hgvs.org/). the strength of evidence in published literature supporting an association between a gene and a disorder, genotype- Data availability ff phenotype correlations, mode(s) of inheritance, and di er- Per institutional review board approval (Western IRB; WIRB ential diagnoses. This curation led to the development of 3 #20161796), all reportable variants identified at Invitae were single-gene tests (DMD, PMP22, and SMN1) and 11 multi- deposited into the ClinVar database.18 The list of genes in each gene panels, with some overlap in genes to address clinical multigene panel is presented in table e-1 (links.lww.com/ heterogeneity (table e-1, links.lww.com/NXG/A246). NXG/A246), and a complete list of variant classifications per gene is listed in table e-2 (links.lww.com/NXG/A247). Next-generation sequencing NGS-based gene panels (not exome-based) were sequenced at high depth coverage (50× minimum, 350× average) to Results simultaneously identify single-nucleotide variants (SNVs), short and long indels, exon-level deletions/duplications, or We performed a diagnostic genetic testing for 25,356 indi- CNVs. Structural rearrangements with breakpoints in coding viduals in aggregate. Because some individuals had multiple

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Table 1 Percentage of orders for each gene panel in which definitive molecular diagnostic (Dx) and variant(s) of uncertain significance in the absence of positive findings (VUS rate) were detected

No. of mol No. of Mol diagnosis diagnosis- No. of diagnosis- VUS (no. Genes with highest Panel orders (no. and rate) CNV sequence variant and yield) diagnostic yield

Cardiomyopathy/ 778 52 (6.68%) 6 46 538 (69.15%) TTN (15.4%), DMD (11.5%), LMNA (7.7%), skeletal muscle TNNT2 (5.8%), DSP (5.8%), MYBPC3 (5.8%)

CMT 1,640 504 (30.73%) 299 205 429 (26.16%) PMP22 (59%), MFN2 (9.7%), GJB1 (8.7%), MPZ (7.5%)

Muscular 903 295 (32.67%) 69 226 342 (37.87%) DMD (62.9%), DYSF (5.9%), dystrophy CAPN3 (5.1%)

Myopathy 1,082 104 (9.61%) 1 103 563 (52.03%) RYR1 (21.3%), TTN (12.4%), ACTA1 (6.7%), DNM2 (5.6%), LMNA (5.6%)

Neuromuscular 5,110 769 (15.05%) 154 615 2,995 (58.61%) DMD (18.3%), RYR1 (8.9%), SMN1 disorders (5.9%), TTN (5.4%), LMNA (5.1%)

Neuropathies 11,302 1,352 (11.96%) 711 641 5,167 (45.72%) PMP22 (52.5%), TTR (7.9%), MPZ (7.4%), MFN2 (6.9%)

Congenital 650 28 (4.31%) 1 27 146 (22.46%) CHRNE (53.6%), RAPSN (17.9%), DOK7 myasthenic (10.7%), CHRNA1 (7.1%), COLQ (7.1%) syndrome

Dystonia 1,910 151 (7.91%) 14 137 225 (11.78%) SGCE (20.5%), TOR1A (19.9%), GCH1 (17.2%), PRRT2 (15.9%), ATP1A3 (9.9%)

Hereditary 2,129 296 (13.9%) 34 262 787 (36.97%) SPAST (43.1%), SPG7 (12.5%), ATL1 spastic (10.4%), SPG11 (9.1%) paraplegia

Limb-girdle 400 89 (22.25%) 12 77 112 (28%) DMD (24.4%), CAPN3 (15.4%), FKRP (11.5%), muscular DYSF (10.3%), ANO5 (7.7%), LMNA (7.7%) dystrophy

Periodic paralysis 1,540 149 (9.68%) 0 149 195 (12.66%) SCN4A (59.7%), CACNA1S (36.2%)

PMP22 365 139 (38.08%) 136 3 4 (1.1%) PMP22 (100%) neuropathy

SMA 3,891 803 (20.64%) 803 0 17 (0.44%) SMN1 (100%)

DMD/BMD 890 327 (36.74%) 227 100 42 (4.72%) DMD (100%)

Abbreviations: BMD = Becker muscular dystrophy; CMT = Charcot-Marie-Tooth; CNV = copy number variant; DMD = Duchenne muscular dystrophy; SMA = spinal muscular atrophy; VUS = variants of uncertain significance. Among positive results, the percentage that are CNVs and sequence variants (single-nucleotide variants) are called out. For each panel, the gene(s) con- tributing to the most positive reports is also given. tests, a combined total of 32,590 analyses were conducted. (CMT), limb-girdle muscular dystrophy, and the compre- The age range of individuals at the time of genetic testing was hensive panel for neuromuscular disorders. Of the 163 genes <1–96 years, with a median of 43 years. Approximately 45% of that provided definitive diagnostic results, PMP22 provided tested individuals were women. More than 90% of the tests the largest number, followed by SMN1, DMD, MFN2, MPZ, were ordered by neurologists or clinical geneticists, and the SPAST, TTR, SCN4A, and GJB1 (figure 2, table 1, and table remaining were ordered by a wide range of other specialists. e-1, links.lww.com/NXG/A246). Notably, 17 individuals in this cohort received molecular diagnoses in 2 genes, and one Definitive diagnoses individual received molecular diagnoses in 3. Most of these Adefinitive molecular diagnosis was obtained for 5,055 of the cases involved pathogenic variants in TTN, MPZ, PMP22, and 25,356 individuals, representing a diagnostic yield of 20% RYR1. No molecular diagnoses were obtained at all from 103 overall and a range of 4%–33%, depending on the panel used genes, most of which are associated with extremely rare (figure 1, table 1, table e-2, links.lww.com/NXG/A247). conditions that have been reported in a handful of individuals Single-gene tests for Charcot-Marie-Tooth disease type 1A and/or in specific ethnic backgrounds. (CMT1A) and Duchenne/Becker muscular dystrophy (DMD/BMD) had 38% and 37% yield, respectively, whereas Classification of variants SMA had a 21% yield. Multigene panels with the highest yield For the 25,356 individuals in this cohort, we reported 33,551 were those for muscular dystrophy, Charcot-Marie-Tooth variants classified as LP/P or VUS. Among these variants, 84%

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 Figure 1 The number of tests ordered and diagnostic yield by panel

Diagnostic yield by panel. The percentage of definitive positive results or MDx for each panel is indicated along the X-axis. Total number of orders is indicated along the Y axis. BMD = Becker muscular dystrophy; CMT = Charcot-Marie-Tooth; DMD = Duchenne muscular dystrophy; HSP = hereditary spastic paraplegia; MD = muscular dystrophy; MDx = molecular diagnostic rate; SMA = spinal muscular atrophy.

were single-nucleotide changes, 6% were indels, and the observed multiple instances of several other common mutations remaining 10% were intragenic or whole gene CNVs. Notably, including 77 instances of GAA c.-32-13T>G, 68 instances of 39% of all positive variants in this clinical cohort (7,789 FIG4 c.122T>C (p.Ile41Thr), 63 instances of FKRP c.826C>A variants) were CNVs, most (93%) of which were in 3 (p.Leu276Ile), 63 instances of TTR c.424 G>A (p.Val142Ile), genes—SMN1, PMP22,andDMD. However, it is important 59 instances of ANO5 c.191dupA (p.Asn64Lysfs*15), 53 that another 77 genes contained 224 clinically significant instances of SPG7 c.1529C>T (p.Ala510Val), and 48 instances – CNVs, accounting for 7% of all clinically significant CNVs of SH3TC3 c.2860C>T (p.Arg954*).19 22 The recently repor- (figure 3, A and B and table 1). Of these, 39 occurred at a low ted intronic variant in COL6A1 (NM_001848.2:c.930+189- frequency and 113 were novel variants. Of the 3,366 VUS/LP/ C>T) was the most common pathogenic variant affecting a type P CNV reported in aggregate, 75% included loss or gain of 6 collagen gene.23,24 a whole gene and the remaining were partial-gene events. However, when restricted to unique CNVs, of which there Testing patterns for commonly were 619, only 18% involved a whole gene and the rest were referred disorders partial-gene events. In other words, many whole gene CNVs For some neuromuscular disorders, the traditional approach were recurrent and often involved SMN1 or PMP22. Further- has been to use single-gene tests rather than gene panels. Al- more, 1,723 of the 3,051 LP/P CNVs observed in this cohort though this may be useful for individuals who present with were in genes associated with AD or XL inheritance and the overt and characteristic phenotypes, those with milder or remaining were in genes associated with AR inheritance. Of the atypical clinical presentations may not attract the requisite at- 1,328 LP/P CNVs observed in AR conditions, 30 were ob- tention and testing. Even in the case of clearly recognizable served as heterozygous events in combination with another disorders, gene panels may be used to curtail a potential di- LP/P non-CNV type of variant and 856 were present in the agnostic odyssey or to avoid invasive procedures such as muscle homozygous state. biopsies. Clinicians without extensive training in neuromus- cular disorders may also be more comfortable with gene panels This cohort revealed several common pathogenic variants that if they can address differential diagnoses broadly. DMD and have been previously described. The SMN1 wholegenedeletion CMT1A are examples of recognizable disorders for which and the PMP22 whole gene duplication and deletion were the many individuals are diagnosed through multigene panels. In most common pathogenic events in this study. We also our cohort, the DMD gene provided positive results for 634

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG Figure 2 Diagnostic yield by gene

The Y axis shows the percentage of positive rate for each gene cumula- tively across all panels it was ordered from, with only the top 30 shown from among the 266 total genes included in panels for neuromuscular disorders.

male patients. Another 146 female individuals were found to be specific panel and 12% using single-gene analysis. Finally, we heterozygous for an LP/P DMD variant; however, inadequate observed that individuals referred for genetic testing for SMA clinical/family history made it impossible to unambiguously received positive diagnostic results most often with single- categorize the indication for testing as either diagnostic or gene analysis (93% of cases). The remaining obtained positive carrier status. Of the 634 males diagnosed with DMD or BMD, results through gene panel testing via the comprehensive 51% underwent single-gene testing, whereas the remaining neuromuscular disorders panel and less often through the underwent multigene testing through a comprehensive mus- comprehensive neuropathies panel. cular dystrophy panel (25%), a limb-girdle muscular dystrophy panel (3%), a comprehensive neuromuscular disorders panel Rare genetic causes (20%), or a cardiomyopathy and skeletal muscle disease panel In more than 6% of individuals for whom a diagnosis of (0.5%). Most individuals (61%) with positive results in DMD a recognizable condition (e.g., DMD, SMA, or CMT1A) had had pathogenic CNVs, and the remaining had SNVs. been established by a referring physician based on clinical features, multigene panel testing revealed a diagnosis in an- We also observed positive results in PMP22 through both other gene related to the phenotypic spectrum of the disorder. single-gene testing and gene panel testing. Most (83%) of the Among 2,501 individuals for whom a clinician received positive results in PMP22 were represented by the classic a negative result for a single-gene or small panel and sub- whole gene duplication associated with CMT1A. Another sequently pursued a larger panel, 200 showed diagnostic 15% were represented by the reciprocal deletion associated results on an expanded panel. In many cases, these results with hereditary neuropathy with pressure palsies (HNPP), were in genes that are rare contributors to neuromuscular and the remaining positive results included 22 SNVs and disease. In relation to DMD, PMP22,orSMN1, specifically, small indels, and one exon 5 deletion. Of the 1,146 individuals when individual analysis of these 3 genes yielded negative who received molecular diagnosis of either CMT1A or results in 339 individuals, a multigene panel provided 57 HNPP, 62% were identified using the comprehensive neu- positive diagnoses that would have otherwise been missed by ropathy panel, whereas 26% were identified using the CMT- a traditional single-gene approach.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 5 Figure 3 Contribution of intragenic CNVs to diagnostic yield by panel

Contribution of intragenic CNVs to di- agnostic yield by panel. (A) The chart shows the percentage of likely patho- genic or pathogenic variants that are se- quence variants (green) or intragenic copy number variants (blue) shown by different panels for neuromuscular dis- orders. Selected cases of intragenic CNVs detected in patients. (B) The baseline (in- dicated by “1.0”) represents the presence of 2 copies of each gene (1 copy of DMD in males). Gray dots indicate data for each internally included control samples. Green dots represent the calculated me- dian of all control samples. Red dots represent data from the patient sample. Data points on the horizontal axis (chro- mosomal coordinates) are clustered at targeted exons. Panels from top to bot- tom show a hemizygous duplication of exons 12-16 of DMD in a male individual, a heterozygous deletion of exons 45-53 of DMD in a female carrier, a heterozy- gous deletion of exons 7 and 8 of MFN2, and a homozygous deletion of exon 1 of SPG7. BMD = Becker muscular dystrophy; CMT = Charcot-Marie-Tooth; CNV = copy number variant; DMD = Duchenne mus- cular dystrophy; HSP = hereditary spastic paraplegia; MD = muscular dystrophy; SMA = spinal muscular atrophy.

We also investigated the number of individuals who received otherwise go undetected if single-gene or small panels were a molecular diagnosis that was not consistent with the referral pursued. For example, among individuals referred with a clini- indication, illuminating differential diagnoses that would cal diagnosis of SMA, most who had a positive molecular

6 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG diagnostic result indeed had the common deletion of SMN1. small cohorts with homogeneous phenotypes rather than However, at least 16 individuals in that group had a molecular a large unselected cohort, as in this study. This and similar diagnosis other than SMN1, with roughly half of these genes studies have not included repeat expansion/contraction dis- associated with muscular dystrophy or myopathies and the orders, the diagnosis of which depends largely on non- – other half with neuropathies. Similarly, among males referred sequencing methods. Although whole exome sequencing,25 27 with an explicit mention of DMD as a suspected clinical di- and targeted exome sequencing,28 can also be used for di- agnosis and for whom we found a positive molecular diagnostic, agnosing neuromuscular disorders, their higher cost, in- three-quarters had a molecular diagnosis in the DMD gene, but complete coverage, long turnaround time to results, and limited the remaining one-quarter had diagnostic variants in other ability to detect intragenic CNVs make them more appropriate genes. Of interest, 19% of these genes were not in a muscular as a second test following a negative panel result. dystrophy panel but instead in a neuromuscular disease panel. Specifically, at least 3 individuals who were suspected to have Most previous studies of the molecular basis of neuromus- DMD were found to have homozygous deletion of SMN1. cular disorders have not routinely included intragenic CNV Other cases were found to have congenital myopathy, cardio- analysis for all genes. We recently showed that intragenic myopathy, congenital myasthenic syndrome, myotonia, or CNVs are important contributors to pathogenic variant bur- spastic paraplegia. den in a broad range of hereditary disorders and should be routinely assessed.15 In this cohort, 39% of all positive results Variants of uncertain significance included CNVs. Over 80% of unique CNVs in our study Although NGS provides the advantage of testing many genes at included only a few exons, emphasizing the need for high once, it also uncovers variants with meager evidence for path- resolution. The majority of pathogenic CNVs were found in ogenicity that therefore get classified as VUS. We surveyed the PMP22, DMD, and SMN1, as expected. Intragenic CNVs VUS in this cohort to determine their distribution and rates of were also identified in 77 other genes for which deletion/ reclassification. In this cohort of 25,356 individuals, the number duplication analysis is not traditionally performed in a single of VUS ranged from 1 to 13 per person (mean, 1.9; median, 1) assay, and these rare CNVs contributed to a molecular di- and 53% of individuals received reports with at least one VUS agnosis in 113 cases. Furthermore, identifying CNVs in genes (table 1). Among the 25,762 VUS reported in this cohort, there with AR inheritance can be particularly helpful because they were 17,321 unique variants distributed across 266 genes. At can exist as compound heterozygous alleles that may be in- least 16% of individuals with one or more VUS had a co- visible to traditional sequencing methods. We confirmed 30 occurring definitive diagnostic result, indicating that the VUS in compound heterozygous diagnoses involving CNVs in genes these cases were likely unrelated to the presenting clinical such as LAMA2, SPG11, DOK7, and PRKN. CNV analysis is phenotype. therefore clearly a necessary component of diagnostic genetic testing for inherited neuromuscular conditions to ensure high After follow-up studies, 2% of these 17,630 unique variants clinical sensitivity. were reclassified, 44% (or 158 variants) changed to LP/P status, and the remaining 198 variants to LB/B status, and Multigene NGS analysis advances the interpretation of het- most of the LP/P reclassifications occurred in genes associ- erogeneity for any single clinical disorder and also helps refine ated with AD disorders and high penetrance. At least 21 genes differential diagnoses. Panels can also be useful for individuals with over 100 unique VUS had 2% or more of VUS resolved, for whom a single-gene test cannot be confidently selected and one gene (DNM2) had over 5% of VUS resolved (figure because of a mild or uncharacteristic phenotype. In illustrating 4). Among the additional evidence that supported reclassifi- how the challenge of genetic heterogeneity can be overcome, cation of a VUS to a clinically significant LP/P result, de novo our data show that many individuals who received a molecular occurrence in genes associated with dominant inheritance was diagnosis for a well-recognized disorder, such as DMD or the most common contributing factor (39%), followed by CMT1A, were actually diagnosed through multigene panels case reports in the published literature (34%) and confirma- rather than single-gene analysis. Moreover, in 2,501 instances tion of trans phase for variants in genes associated with AR in which a clinician received a negative result for a single-gene inheritance (10%). or small panel test and subsequently pursued testing using a larger panel, a positive diagnostic result was obtained for 200 individuals. Separately, NGS panels also help address differ- Discussion ential diagnoses by supporting evaluation of genes for related disorders. Several individuals with clinical suspicion of DMD Our study has demonstrated that NGS-based gene panels with or CMT1A received positive molecular diagnoses in genes simultaneous sequence and CNV detection can provide a di- unrelated to their referral indication. For example, 133 indi- agnostic yield of 4%–33% in individuals with neuromuscular viduals suspected to have DMD instead had a molecular di- disease, including muscular dystrophies, SMA, spastic para- agnosis in a gene unrelated to muscular dystrophy but related plegia, neuropathies, congenital myasthenic syndromes, and to one of the several other types of neuromuscular disorders, congenital myopathies. Although higher diagnostic rates have including spastic paraplegia, congenital myasthenic syn- been published in other studies, they have typically involved drome, and SMA.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 7 Figure 4 Distribution of variants of uncertain significance and rate of reclassification across genes

Distribution of variants of uncertain significance and rate of reclassification across genes (left axis). The gray bars indicate the number of unique variants (bottom axis) that were clas- sified as variants of uncertain significance in each gene and the red bars indicate the percentage (top axis) that were reclassified to likely pathogenic or pathogenic or to likely benign or benign. Only genes with >100 unique VUS are in- cluded. Most de novo events were observed in RYR1, ACTA1, SPAST, and MPZ. Most downgrades of VUS to likely benign or benign classifications occurred in KBTBD13, PLEC, NEB, and LAMA2. VUS = variants of uncertain significance.

Our study also provided insight into the complexities of were in genes associated with AR inheritance for which interpreting rare variants observed in the many genes asso- demonstration of trans phase provided useful evidence to- ciated with neuromuscular disorders. Investigation of VUS in ward pathogenicity. this cohort showed that most occurred as single heterozygous alleles in genes associated with AR disorders and were This study has better illuminated the diagnostic yields for therefore less likely to be disease-causing. By contrast, VUS in various neuromuscular disorders, the contribution of CNVs genes associated with AD disorders and high penetrance have to these disorders, the importance of addressing genetic a higher likelihood of being reclassified to a clinically signifi- heterogeneity and differential diagnoses, and the occurrence cant status following studies to investigate segregation or de and resolution of VUS. As NGS becomes even more acces- novo status. Of the VUS that were resolved to clinically sig- sible to individuals with neuromuscular disorders, these types nificant results, 48% were in the genes associated with AD of studies will provide clinicians the requisite information for inheritance and were found to be de novo or, alternatively, understanding the utility of genetic tests, establishing

8 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG expectations for affected individuals in their course of care, 2. Dohrn MF, Gl¨ockle N, Mulahasanovic L, et al. Frequent genes in rare diseases: panel- based next generation sequencing to disclose causal mutations in hereditary neu- and managing test results. ropathies. J Neurochem 2017;143:507–522. 3. Nigro V, Savarese M. Next-generation sequencing approaches for the diagnosis of skeletal muscle disorders. Curr Opin Neurol 2016;29:621–627. Study funding 4. Reale C, Panteghini C, Carecchio M, Garavaglia B. The relevance of gene panels in No targeted funding reported. movement disorders diagnosis: a lab perspective. Eur J Paediatr Neurol 2018;22: 285–291. 5. B¨onnemann CG, Wang CH, Quijano-Roy S, et al. Diagnostic approach to the con- Disclosure genital muscular dystrophies. Neuromuscul Disord 2014;24:289–311. Disclosures available: Neurology.org/NG. 6. Sansone VA. The dystrophic and nondystrophic myotonias. Continuum 2016;22: 1889–1915. 7. Boutry M, Morais S, Stevanin G. Update on the genetics of spastic paraplegias. Curr Publication history Neurol Neurosci Rep 2019;19:18. 8. Prior R, Van Helleputte L, Benoy V, Van Den Bosch L. Defective axonal transport: Received by Neurology: Genetics May 23, 2019. Accepted in final form a common pathological mechanism in inherited and acquired peripheral neuropa- December 30, 2019. thies. Neurobiol Dis 2017;105:300–320. 9. Nallamilli BRR, Chakravorty S, Kesari A, et al. Genetic landscape and novel disease mechanisms from a large LGMD cohort of 4656 patients. Ann Clin Transl Neurol 2018;5:1574–1587. 10. Iqbal Z, Rydning SL, Wedding IM, et al. Targeted high throughput sequencing in Appendix Authors hereditary ataxia and spastic paraplegia. PLoS One 2017;12:e0174667. 11. Tian X, Liang WC, Feng Y, et al. Expanding genotype/phenotype of neuromuscular Name Location Contribution diseases by comprehensive target capture/NGS. Neurol Genet 2015;1:e14. doi: 10. 1212/NXG.0000000000000015. Thomas L. Invitae Designed and conceptualized the 12. DiVincenzo C, Elzinga CD, Medeiros AC, et al. The allelic spectrum of Charcot- Winder, PhD study; analyzed the data; drafted the Marie-Tooth disease in over 17,000 individuals with neuropathy. Mol Genet Genomic manuscript for intellectual content Med 2014;2:522–529. 13. Stehl´ıkov´a K, Sk´alov´aD,Z´ıdkov´a J, et al. Muscular dystrophies and myopathies: the Christopher A. Invitae Analyzed the data; drafted the spectrum of mutated genes in the Czech Republic. Clin Genet 2017;91:463–469. Tan, MS manuscript 14. Chae JH, Vasta V, Cho A, et al. Utility of next generation sequencing in genetic diagnosis of early onset neuromuscular disorders. J Med Genet 2015;52:208–216. Sarah Klemm, Invitae Analyzed the data; drafted the 15. Truty R, Paul J, Kennemer M, et al. Prevalence and properties of intragenic copy- MS manuscript number variation in Mendelian disease genes. Genet Med 2019;21:114–123. 16. Lincoln SE, Kobayashi Y, Anderson MJ, et al. A systematic comparison of traditional Hannah White, Invitae Analyzed the data; drafted the and multigene panel testing for hereditary breast and ovarian cancer genes in more – MS manuscript than 1000 patients. J Mol Diagn 2015;17:533 544. 17. Nykamp K, Anderson M, Powers M, et al. Sherloc: a comprehensive refinement of the fi – Jody M. Invitae Analyzed the data; drafted the ACMG-AMP variant classi cation criteria. Genet Med 2017;19:1105 1117. fi Westbrook, manuscript 18. Landrum MJ, Kattman BL. ClinVar at ve years: delivering on the promise. Hum – PhD Mutat 2018;39:1623 1630. 19. Brockington M, Yuva Y, Prandini P, et al. Mutations in the fukutin-related James Z. Wang, Invitae Analyzed the data; drafted the protein gene (FKRP) identify limb girdle muscular dystrophy 2I as a milder PhD manuscript allelic variant of congenital muscular dystrophy MDC1C. Hum Mol Genet 2001; 10:2851–2859. Ali Entezam, Invitae Analyzed the data; drafted the 20. Høyer H, Braathen GJ, Busk Ø, et al. Genetic diagnosis of Charcot-Marie-Tooth PhD manuscript disease in a population by next-generation sequencing. Biomed Res Int 2014;2014: 210401. Rebecca Truty, Invitae Principal data acquisition and 21. Roxburgh RH, Marquis-Nicholson R, Ashton F, et al. The p.Ala510Val mutation PhD analysis in the SPG7 (paraplegin) gene is the most common mutation causing adult onset neurogenetic disease in patients of British ancestry. J Neurol 2013;260: – Robert L. Invitae and Revised the manuscript for 1286 1294. Nussbaum, MD UC-San intellectual content 22. Bouquet F, Coss´ee M, B´ehin A, et al. Miyoshi-like distal myopathy with mutations in – Francisco anoctamin 5 gene. Rev Neurol (Paris) 2012;168:135 141. 23. Cummings BB, Marshall JL, Tukiainen T, et al. Improving genetic diagnosis in Elizabeth M. Northwestern Revised the manuscript for Mendelian disease with transcriptome sequencing. Sci Transl Med 2017;9. 24. Bolduc V, Foley AR, Solomon-Degefa H, et al. A recurrent COL6A1 pseudoexon McNally, MD, University intellectual content ff PhD insertion causes muscular dystrophy and is e ectively targeted by splice-correction therapies. JCI Insight 2019;4:e124403. 25. Waldrop MA, Pastore M, Schrader R, et al. Diagnostic utility of whole exome se- Swaroop Invitae Designed and conceptualized the quencing in the neuromuscular clinic. Neuropediatrics 2019;50:96–102. Aradhya, PhD study; interpreted the data; revised 26. Ghaoui R, Cooper ST, Lek M, et al. Use of whole-exome sequencing for diagnosis of the manuscript for intellectual limb-girdle muscular dystrophy: outcomes and lessons learned. JAMA Neurol 2015; content 72:1424–1432. 27. Reddy H, Cho K-H, Lek M, et al. The sensitivity of exome sequencing in identifying pathogenic mutations for LGMD in the United States. J Hum Genet 2017;62: 243–252. References 28. Johnson K, Bertoli M, Phillips L, et al. Detection of variants in dystroglycanopathy- 1. Biancalana V, Laporte J. Diagnostic use of massively parallel sequencing in neuro- associated genes through the application of targeted whole-exome sequencing analysis muscular diseases: towards an integrated diagnosis. J Neuromuscul Dis 2015;2: to a large cohort of patients with unexplained limb-girdle muscle weakness. Skelet 193–203. Muscle 2018;8:23.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 9 CLINICAL/SCIENTIFIC NOTES OPEN ACCESS Gerstmann-Str¨aussler-Scheinker disease (PRNP p.D202N) presenting with atypical parkinsonism

Simone Baiardi, MD, Romana Rizzi, MD, PhD, Sabina Capellari, MD, Anna Bartoletti-Stella, PhD, Correspondence Andrea Zangrandi, MSc, Federico Gasparini, MSc, Enrico Ghidoni, MD, and Piero Parchi, MD, PhD Dr. Parchi [email protected] Neurol Genet 2020;6:e400. doi:10.1212/NXG.0000000000000400

The p.D202N mutation in PRNP is a rare variant associated with Gerstmann-Str¨aussler- Scheinker disease (GSS), a genetic form of prion cerebral amyloidosis. To date, there have been only 4 reports of this mutation worldwide and only one detailed clinical study (table e-1, links. – lww.com/NXG/A223).1 4 Here, we describe the clinical phenotype and the results of neu- roradiologic and laboratory investigations in an Italian patient carrying this genetic variant.

Case report A 59-year-old Caucasian man, with no family history of neurologic disorders, presented with a 2-year history of slurred speech and progressive gait difficulties causing motor slowing and accidental falls. In the same period, he lost 14 kg and developed urinary incontinence. His medical history was relevant to hypertension. Neurologic examination revealed ideomotor slowing, smooth pursuit deficit, dysarthria, dysmetria, axial and limb plastic hypertonia, brisk deep tendon reflexes, Babinski sign, and ataxic gait. Given the clinical findings, the suspicion of multiple system atrophy (MSA) was raised. Repeated neuropsychological testing revealed abnormalities in at- tention, constructive praxis, lexical access, and verbal memory span, consistent with a multiple domain cognitive impairment (table e-2, links.lww.com/NXG/A223). 123I-ioflupane single- photon emission computerized tomography (SPECT) (DaTscan) showed reduced presynaptic dopamine transporter uptake in the right caudate and putamen bilaterally (figure 1A). Brain MRI displayed multiple, small hyperintense foci in subcortical white matter and right putamen in T2 sequences, whereas diffusion-weighted imaging was negative (figure 1B). Spinal MRI was un- remarkable, except for a tiny C5-C6 disc protrusion (figure 1C). Somatosensory-evoked potentials revealed symmetrically delayed spinal conduction and a reduced motor response from the right lower leg. Electroneurography and sympathetic skin response were normal. EEG showed discharges of slow waves, occasionally pseudorhythmic and sharp, favored by drowsiness but lacking a definite periodism (figure 1D). CSF analyses revealed a positive 14-3-3 protein assay, markedly increased total-tau (3,617 pg/mL, n.v. 44–298) and phosphorylated-tau (337 pg/mL, n.v. 35–66), and normal amyloid-beta 1–42 (982 pg/mL, n.v. 562–1,018). The CSF prion real- time quaking-induced conversion (RT-QuIC) assay was negative. Direct sequencing of the PRNP open reading frame revealed a point mutation at codon 202 (p.D202N), causing the substitution of aspartic acid for asparagine (figure e-1, links.lww.com/NXG/A223) and valine homozygosity at codon 129. The patient lost the walking ability 3 years after the clinical onset; at this time, the patient developed dysphagia, and his speech became unintelligible because of severe dysarthria, but comprehension was relatively spared. The patient died 4.5 years after the onset because of sepsis complications. An autopsy was not performed.

From the Department of Biomedical and Neuromotor Sciences (S.B., S.C.), University of Bologna; IRCCS Istituto delle Scienze Neurologiche di Bologna (S.B., S.C., A.B.-S., P.P.); Neurology Unit (R.R.), Department of Neuro-Motor Diseases, Azienda Unit`a Sanitaria Locale – IRCCS; Clinical Neuropsychology (A.Z., F.G., E.G.), Cognitive Disorders and Dyslexia Unit, Department of Neuro-Motor Diseases, Azienda Unit`a Sanitaria Locale – IRCCS, Reggio Emilia; and Department of Diagnostic Experimental and Specialty Medicine (DIMES) (P.P.), University of Bologna, Italy.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by Azienda USL di Bologna.

Ethical standards: All clinical studies have been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments. The patient signed written informed consent for genetic testing and CSF collection. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Figure 1 Neurophysiologic and neuroimaging findings

(A) 123I-ioflupane SPECT (DaTscan) showing an abnormally reduced uptake in the right caudate nucleus and both putamen nuclei. (B) A few small, focal hyperintensities in the frontal subcortical white matter on brain MRI fluid-attenuated inversion recovery sequence. (C) Spinal MRI demonstrating a tiny C5-C6 disc protrusion. (D) EEG recording showing some bursts of diffuse slow waves (left), promoted by drowsiness (right). R = right, L = left; P = posterior.

The causative role of the p.D202N mutation in our patient is Discussion – supported by the similar presentation among reported cases.2 4 GSS is a genetic prion disease caused by several point and the neuropathologic studies documenting a genuine GSS 5 (i.e., missense, nonsense) or insertional mutations in PRNP. phenotype in 3 previously described cases carrying the The clinical phenotype of GSS is heterogeneous and may in- p.D202N mutation examined neuropathologically, a pheno- clude ataxia, spastic paraparesis, cognitive decline, and amyo- 5 type which has never been reported to date in a subject carrying trophy as presenting signs. Parkinsonism may also occur, but – the wild-type PRNP sequence.1 3 However, given the absence not as the dominant clinical feature, especially at disease onset. of a positive family history in the present case (as well in As the most significant exception, patients with GSS carrying 4 fi the p.D202N substitution consistently showed early extrapyra- others), we cannot de nitely rule out the possibility of an midal features, often raising the suspicion of atypical parkinso- incomplete penetrance of the p.D202N variant. nian syndrome. Notably, family history was strongly positive for parkinsonism in one case (i.e., overall 9 subjects in 5 gen- It is well established that the methionine (M)/valine(V) erations), with the proband’s mother and cousin being di- polymorphism at PRNP codon 129 strongly modulates the 6 agnosedwithprogressivesupranuclearpalsy.2,4 In line with disease phenotype of human prion disease. In GSS-p.D202N, these observations, 123I-ioflupane SPECT (DaTscan) gave ab- the mutation cosegregated with V129. Indeed, valine homo- normal results in both patients who underwent this test (table zygosity at codon 129 was documented in 3 patients, and the e-1, links.lww.com/NXG/A223). In our patient, parkinsonism mutation was in cis with valine in a fourth case (table e-1, links. was not either dominant or isolated at disease onset, but the lww.com/NXG/A223). concurrence of parkinsonism with ataxia and pyramidal signs raised the suspicion of MSA, which belongs to the spectrum of Of interest, the CSF profile in our cases indicated a severe atypical parkinsonisms. neurodegenerative process associated with the accumulation of

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG phosphorylated tau protein, without a concomitant A-Beta protein deposition.7 This finding is also in line with the di- Appendix (continued) fi agnosis of GSS, given that the secondary tau-positive neuro - Name Location Role Contribution brillary pathology is a neuropathologic hallmark of GSS5,e1 and also occurred in the 3 previously described cases carrying the Anna IRCCS Institute of Author Major role in the 1–3 Bartoletti- Neurological Sciences, acquisition of data; p.D202N variant. Finally, the negative CSF prion RT-QuIC Stella, PhD Bologna, IT analyzed and result is not surprising because, despite the few cases analyzed, interpreted the data the suboptimal diagnostic accuracy of this assay for GSS has Andrea Azienda Unit`a Author Major role in the e2-e4 been already reported by different groups. Zangrandi, Sanitaria Locale – acquisition of data; MD IRCCS, Reggio Emilia, analyzed and Italy interpreted the data Our and previous observations underline the importance of ff Federico Azienda Unit`a Author Major role in the considering the GSS-p.D202N-V129 variant in the di erential Gasparini, Sanitaria Locale – acquisition of data; diagnosis of patients with atypical parkinsonism, especially when MD IRCCS, Reggio Emilia, analyzed and associated with cerebellar and/or pyramidal signs and dementia. Italy interpreted the data

Study funding Enrico Azienda Unit`a Author Major role in the Ghidoni, Sanitaria Locale – acquisition of data; No targeted funding reported. MD IRCCS, Reggio Emilia, analyzed and Italy interpreted the data

Disclosure Piero University of Bologna; Author Designed and Disclosures available: Neurology.org/NG. Parchi, MD, IRCCS Institute of conceptualized the PhD Neurological Sciences, study; major role in Bologna, IT the acquisition of data; analyzed and Publication history interpreted Received by Neurology: Genetics July 17, 2019. Accepted in final form the data, drafted December 10, 2019. and revised the manuscript for intellectual content Appendix Authors

Name Location Role Contribution

Simone University of Bologna; Author Analyzed and References Baiardi, IRCCS Institute of interpreted the data; 1. Piccardo P, Dlouhy SR, Lievens PM, et al. Phenotypic variability of Gerstmann- MD Neurological Sciences, drafted the Str¨aussler-Scheinker disease is associated with prion protein heterogeneity. Bologna, IT manuscript for J Neuropathol Exp Neurol 1998;57:979–988. intellectual content 2. Kong Q, Surewicz WK, Petersen RB, et al. Inherited prion diseases. In: Prusiner SB, Prion Biology and Diseases. 2nd ed. Woodbury: Cold Spring Harbor Laboratory Romana Azienda Unit`a Author Major role in the Press; 2004:673–776. Rizzi, MD, Sanitaria Locale – acquisition of data; 3. Fleming AB, Ghetti B, Murrell IR, et al. Gerstmann-Str¨aussler-Scheinker disease with PhD IRCCS, Reggio Emilia, analyzed and progressive supranuclear palsy presentation. Dement Geriatr Cogn Disord 2010;30(1 Italy interpreted the data; suppl):43. revised the 4. Plate A, Benninghoff J, Jansen GH, et al. Atypical parkinsonism due to a D202N manuscript for Gerstmann-Str¨aussler-Scheinker prion protein mutation: first in vivo diagnosed case. intellectual content Mov Disord 2013;28:241–244. 5. Ghetti B, Piccardo P, Zanusso G. Dominantly inherited prion protein cerebral Sabina University of Bologna; Author Major role in the amyloidoses—a modern view of Gerstmann-Str¨aussler-Scheinker. Handb Clin Capellari, IRCCS Institute of acquisition of data; Neurol 2018;153:243–269. MD Neurological Sciences, analyzed and 6. Baiardi S, Rossi M, Capellari S, Parchi P. Recent advances in the histo-molecular Bologna, IT interpreted the data; pathology of human prion disease. Brain Pathol 2019;29:278–300. revised the 7. Jack CR Jr, Bennett DA, Blennow K, et al. A/T/N: an unbiased descriptive classi- manuscript for fication scheme for Alzheimer disease biomarkers. Neurology 2016;87:539–547. intellectual content Data available from supplement data (References e1–e4): links.lww.com/NXG/A223

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 CLINICAL/SCIENTIFIC NOTES OPEN ACCESS 4H leukodystrophy Mild clinical phenotype and comorbidity with multiple sclerosis

Stephanie M. DeGasperis, MSc, Genevi`eve Bernard, MD, Nicole I. Wolf, MD, PhD, Elka Miller, MD, and Correspondence Daniela Pohl, MD, PhD Dr. Pohl [email protected] Neurol Genet 2020;6:e409. doi:10.1212/NXG.0000000000000409

Hypomyelinating leukodystrophy with hypodontia and hypogonadotropic hypogonadism (4H leukodystrophy), also known as POLR3-related leukodystrophy, is a genetic disorder caused by autosomal recessive mutations in the POLR3A, POLR3B, POLR1C,orPOLR3K – genes.1 3 Most patients have progressive motor deficits.4 We present 2 siblings with a milder phenotype and lack of disease progression previously reported within a larger cohort of patients.5

Case presentation Case 1 A 16-year-old previously healthy girl was referred to our neurology clinic with a 3-week history of numbness in her lower extremities, headaches, and blurred vision in her right eye. On examination, she had ataxic gait because of decreased sensation to touch in her lower ex- tremities and hyperreflexia. MRI showed enhancement of the right optic nerve, diffuse sym- metric signal abnormalities in the white matter of both cerebral hemispheres, and multiple spinal cord lesions (figure). The CSF analysis revealed oligoclonal bands. She was diagnosed with right optic neuritis and a hypomyelinating leukodystrophy.

She experienced 3 demyelinating relapses. The first occurred 1 month after her initial visit and presented as left optic neuritis. She was treated with monthly IV immunoglobulin for 18 months. At age 18 years, she experienced her second relapse with leg weakness and fatigue. MRI revealed new hyperintense lesions in brain areas and the spinal cord that were not present 7 months before (figure). Her care was transferred to adult neurology. She was diagnosed with multiple sclerosis (MS) and treatment with interferon-β-1a was initiated. Her third relapse, at age 21 years, again involved bilateral leg weakness. A repeat MRI showed no progression of her disease.

At age 21 years, genetic testing revealed that she was homozygous for the POLR3B pathogenic variant V523E, and she was diagnosed with 4H leukodystrophy. The patient continued to have no clinical symptoms of 4H leukodystrophy other than myopia. Her neurologic examination and cognition were normal. She reported suffering from some anxiety and depression.

At the last clinical follow-up at age 26 years, she worked a full-time job. In view of her mood disorder, she had been switched to an immunomodulatory treatment with glatiramer acetate. She had not experienced any progression of her symptoms and had no additional MS relapses.

From the Faculty of Medicine (S.M.D., D.P.), University of Ottawa, ON, Canada; Departments of Neurology and Neurosurgery, Pediatrics and Human Genetics (G.B.), McGill University; Department Specialized Medicine (G.B.), Division of Medical Genetics, McGill University Health Center; Child Health and Human Development Program (G.B.), Research Institute of the McGill University Health Center; MyeliNeuroGene Laboratory (G.B.), Research Institute of the McGill University Health Center, Montreal, Quebec, Canada; De- partment of Pediatric Neurology (N.I.W.), Emma Children’s Hospital, Amsterdam, Netherlands; Amsterdam Neuroscience (N.I.W.), Vrije Universiteit, Netherlands; and Department of Medical Imaging (E.M.) and Division of Neurology (D.P.), CHEO, University of Ottawa, ON, Canada.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Figure MRI images of a pair of siblings with patterns of hypomyelination

Patient 1 (A–J): MRI at 16 years of age (A–E). Axial T2 (A and B), sagittal T2 (C and E), and sagittal T1 with contrast (D). Images demonstrate T2 hyperintensity of the white matter in keeping with hypomyelination (* in A and B) with relative preservation of myelin in the anterior limb of the internal capsules, the corpus callosum, optic radiations (A), corticospinal tracts up to the apex (B). Demyelinating lesions within the spinal cord (C–E). MRI at 18 years of age (F–J). Sagittal T2 (F), sagittal T1 without contrast (G), sagittal T2 (H), and axial T2 (I and J). Enlargement of the spinal cord lesion (arrow in F). T1 hypointense and T2 hyperintense oval white matter lesion (arrows in G and H), along the path of the deep medullary vein (Dawson finger, arrows in H and I). Ovoid lesion of the right centrum semiovale (arrow in J). The corpus callosum is thin (F). Patient 2 (K–M): MRI at 16 years of age. Axial T2 (K and L) and sagittal T1 (M). Hypomyelination (* in K and L) and less residual myelin compared with patient 1, but more residual myelin than is typically seen in 4H leukodystrophy. No white matter lesions in thebrain or spinal cord (M). The corpus callosum is thin (M).

Case 2 his sister was also found in his consanguineous parents who The younger brother of the first patient was referred to the were heterozygous carriers. Both parents had short stature but neurology clinic at age 15 years after his first focal-to-bilateral were otherwise asymptomatic with normal dentition, endo- tonic-clonic seizure. His medical history was unremarkable, crine, cognition and neurologic status. The MRIs of both apart from a learning disability diagnosed at age 11 years. His parents were normal. neurologic examination was normal with the exception of some fi stumbling on tandem gait testing and myopia. His EEG showed Repeat MRI of the patient 1 year after his rst presentation was evidence of epileptiform discharges predominantly in the left unchanged. He had no neurologic complaints. On examination, fl posterior region and intermittently slow background activity. he had abnormal upgaze saccades, hyperre exia in the lower extremities (3+), and mild dysmetria with heel-to-shin testing. At age 16 years, he had a second presumably bilateral tonic- clonic seizure (unwitnessed, parents found him postictal) and On his last follow-up, the patient was 23 years old. He had was administered carbamazepine. His MRI showed bilateral remained seizure free on carbamazepine for 5 years and did not have new neurologic deficits. diffuse, symmetric signal changes in the white matter com- patible with hypomyelination, thinning of the corpus cal- losum, and normal myelination of the spinal cord (figure). Discussion At age 18 years, the diagnosis of 4H leukodystrophy was The patients discussed have atypical presentations of 4H confirmed with genetic testing. The same V523E mutation as in leukodystrophy. Most reported patients with 4H

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG leukodystrophy have progressive neurologic deficits, with or without dental abnormalities, endocrine problems, and/or Appendix Authors 2,5,6 myopia. Typical clinical presentations range from early Name Location Contribution childhood motor clumsiness to cognitive difficulties appar- ent from later childhood.2,5,6 Patients’ brain MRIs show Stephanie M. University of Ottawa, Analyzed the data; drafted DeGasperis, ON, Canada the manuscript for diffuse hypomyelination with relative preservation of specific MSc intellectual content structures. Our patients have significantly more residual Genevi`eve McGill University, Revised the manuscript for myelin (figure). Patients with POLR3B mutations typically Bernard, MD Montreal, Quebec, intellectual content present slightly earlier but have a milder disease course than Canada 5,7 those with POLR3A mutations. It is unusual for a 4H Elka Miller, CHEO, University of Interpreted the data; revised leukodystrophy patient to have absent neurologic deficits by MD Ottawa, ON, Canada the manuscript for age 26 years. intellectual content Nicole I. Wolf, Emma Children’s Revised the manuscript for MD, PhD Hospital, Amsterdam, intellectual content It is common for patients to have one copy of the V523E Netherlands mutation in the POLR3B gene; however, homozygosity is Daniela Pohl, CHEO, Research Designed the study; revised rare. We hypothesize that patients carrying 2 copies of this MD, PhD Institute, Ottawa, ON, the manuscript for variant are either normal their entire life or have only mild Canada intellectual content neurologic deficits and therefore do not seek medical atten- tion. Because no other patient with comorbidity of MS and 4H leukodystrophy has been described, we hypothesize that it is coincidental in our patient. References 1. Timmons M, Tsokos M, Asab MA, et al. Peripheral and central hypomyelination with hypogonadotropic hypogonadism and hypodontia. Neurology 2006;67:2066–2069. Study funding 2. Bernard G, Vanderver A. POLR3-related leukodystrophy. In: GeneReviews. Seattle, WA: University of Washington, Seattle; 2012. No targeted funding reported. 3. Dorboz I, Dumay-Odelot H, Boussaid K, et al. Mutation in POLR3K causes hypo- myelinating leukodystrophy and abnormal ribosomal RNA regulation. Neurol Genet 2018;4:e289. doi: 10.1212/NXG.0000000000000289. Disclosure 4. Battini R, Bertelloni S, Astrea G, et al. Longitudinal follow up of a boy affected by Pol III- Disclosures available: Neurology.org/NG. related leukodystrophy: a detailed phenotype description. BMC Med Genet 2015;16:53. 5. Wolf NI, Vanderver A, van Spaendonk RM, et al. Clinical spectrum of 4H leukodys- trophy caused by POLR3A and POLR3B mutations. Neurology 2014;83:1898–1905. Publication history 6. Vanderver A, Tonduti D, Bernard G, et al. More than hypomyelination in Pol-III – fi disorder. J Neuropathol Exp Neurol 2013;72:67 75. Received by Neurology: Genetics June 25, 2019. Accepted in nal form 7. Tewari VV, Mehta R, Sreedhar CM, et al. A novel homozygous mutation in POLR3A December 18, 2019. gene causing 4H syndrome: a case report. BMC Pediatr 2018;18:126.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 CLINICAL/SCIENTIFIC NOTES OPEN ACCESS Differential subcellular expression of P525LFUS as a putative biomarker for ALS phenoconversion

Maria Caputo, MD, Vincenzo La Bella, MD, PhD, and Antonietta Notaro, PhD Correspondence Dr. La Bella Neurol Genet 2020;6:e410. doi:10.1212/NXG.0000000000000410 [email protected]

P525LFused-in-Sarcoma (FUS) mutation is associated with a specific amyotrophic lateral sclerosis (ALS) phenotype characterized by a juvenile-onset and a severe course.1 This harmful point mutation is located in the nuclear localization signal (NLS) domain at the protein C-terminal.2 Although wild-type FUS protein is expressed almost exclusively in the nucleus, the P525LFUS mutation leads to a protein mislocalization into the cytoplasm3,4 because of its loss of capacity to bind its transporter karyopherin-2 and to be transferred back to the nucleus.3

Here, we compare FUS expression and localization in skin fibroblasts of 2 sisters, both carriers of a P525LFUS mutation, belonging to a Sicilian family with this mutation.4,5

The first sister (DC) was seen at age 21 years when she was asymptomatic. After almost 2-year follow-up, she developed a bulbar form of ALS and died 13 months after disease onset. In both conditions, i.e., asymptomatic (DC-A, at the time of the first visit) and symptomatic (DC-S, soon after disease onset), skin fibroblasts were purified and cultured. The other sister (DL) was seen when she was aged 25 years; on that occasion, fibroblasts were also purified. She is at present asymptomatic.

We studied the expression and subcellular localization of FUS protein in fibroblasts from the 2 P525LFUS carriers. Concerning DC, we analyzed FUS expression in the fibroblasts when she was asymptomatic and after the disease onset. A patient with sporadic clinically definite ALS with no known ALS-related gene mutations (sporadic ALS [S-ALS]) and a healthy control (HC) were used as controls.

All individuals and patients involved in this study signed informed consent for the genetic testing and the skin biopsy. The experimental protocol was approved by the Ethics Committee Palermo 1 (July 2017).

Fibroblasts were cultured in Dulbecco’s Modified Eagle Medium supplemented with 10% calf serum and antibiotic/antimycotic solution. Cells were plated on glass coverslips to perform immunofluorescence by using a polyclonal FUS antibody (11570-1-AP, Proteintech Group, Chicago, IL). The analysis of subcellular FUS expression was made through a Zeiss LSM5 confocal microscope. Cell counting for subcellular FUS expression was performed as described by Lo Bello et al.4

As expected, FUS mislocalized to the cytoplasm in almost all fibroblasts carrying the P525LFUS mutation.4 Conversely, control fibroblasts (S-ALS and HC) expressed FUS only in the nucleus (figure, A).

After a more careful inspection, we observed important differences in the nucleus- cytoplasm distribution of FUS protein between the 2 FUS mutants (DC and DL). By visual

From the ALS Clinical Research Center and Laboratory of Neurochemistry, Department of Biomedicine Neuroscience and Advanced Diagnostics (Bi.N.D.), University of Palermo, Italy.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

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.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Figure Expression and differential subcellular distribution of P525LFUS in fibroblasts of asymptomatic and symptomatic mutation carriers

(A) Representative confocal images of immunofluorescence experiments performed with a polyclonal anti-FUS antibody on fibroblasts from the P525LFUS mutation carrier DC, either in the asymptomatic (DC-A) or symptomatic (DC-S) stage, the asymptomatic P525LFUS mutation carrier sister (DL), a patient with sporadic ALS (S-ALS), and HC. Note that while in the P525LFUS mutation carriers, the protein is mislocalized to the cytoplasm; in sALS and HC, FUS remains almost exclusively nuclear. Bar, 10 μm. (B) Percent of FUS expression in the nucleus, in the cytoplasm, or in both the nucleus and the cytoplasm of fibroblasts from the P525LFUS mutation carriers, sALS and HC. Data are expressed as percentage of the total counted cells (mean ± SD of 2 separate experiments performed in duplicate wells). ap < 0.05, percent of cells with FUS expression exclusively in the cytoplasm vs cells with FUS expression in both the nucleus and the cytoplasm from DC-A, DC-S, and DC-L. One-way analysis-of-variance with a post hoc Holm-Sidak analysis. FUS = fused-in-sarcoma; HC = healthy control; sALS = sporadic amyotrophic lateral sclerosis.

counting, most DL fibroblasts showed a combined nuclear might suggest a change from a stable asymptomatic phase and cytoplasmic FUS localization. Exclusive cytoplasmic to incoming disease onset. Therefore, we hypothesize that staining was instead seen in over 35% of fibroblasts from although asymptomatic, DC could have been in a no-return DC in her asymptomatic stage (DC-A), which increased to point already 2 years before clinical onset. 70% after the disease onset (DC-S, figure, B). Thus, a higher number of cells with an exclusively FUS cyto- It would be interesting to verify whether other FUS muta- plasmic localization seems to be related to the phenotype tions, especially those in the NLS at the C-terminal, show conversion. a similar subcellular redistribution after disease onset.

A question arises about the meaning of the reduced nuclear Discussion FUS expression in mutant P525LFUS cells near to and after the disease onset. This abnormal subcellular FUS redistribution This study confirms that in P525LFUS fibroblasts, taken from might express a loss of its nuclear physiologic function; this asymptomatic mutation carriers, the protein is mislocalized to 3,4 might in turn contribute to motoneuron dysfunction and thus the cytoplasm. However, it also shows that after pheno- to the disease onset.6 conversion, in the large majority of mutant cells the protein fi disappear from the nucleus. This occurred to DC broblasts, We suggest that the lack of FUS expression in the nucleus of whose number of cells expressing FUS solely in the cytoplasm fibroblasts of asymptomatic P525LFUS mutation carriers might doubled from some 35% in the asymptomatic phase (DC-A) signal an incipient disease onset, being, therefore, a specific to over 70% after ALS onset (DC-S). The 2 sisters were biomarker of phenoconversion. Our report also highlights the biopsied at the same time when they were in their asymp- importance of the skin changes as representative of concur- tomatic phase. Two years after the first biopsy, when first rent neuronal/glial biological modifications occurring in the symptoms of ALS appeared to DC, DL was still asymptom- disease.7 atic with almost all cells showing a nucleocytoplasmic FUS localization. Author contributions M. Caputo was involved in study concept, data analysis, and The presence of an appreciable number of cells from DC-A writing of the manuscript. V. La Bella was involved in study showing only cytoplasmic FUS expression is intriguing, as it concept, data analysis, critical revision of the manuscript, and

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG study supervision. A. Notaro was involved in writing of the References manuscript and study supervision. 1. Conte A, Lattante S, Zollino M, et al. P525L FUS mutation is consistently associated with a severe form of juvenile amyotrophic lateral sclerosis. Neuromuscul Disord 2012; 22:73–75. Study funding 2. Kwiatkowski TJ Jr, Bosco DA, Leclerc AL, et al. Mutations in the FUS/TLS gene on chromosome 16 cause familial amyotrophic lateral sclerosis. Science 2009;323: This study was in part supported by a liberal donation from 1205–1208. the Italian Association of Amyotrophic Lateral Sclerosis 3. Dormann D, Rodde R, Eddbauer D, et al. ALS-associated fused in sarcoma (FUS) mutations disrupt transportin-mediated nuclear import. EMBO J 2010;29:2841–2857. (AISLA), Palermo Unit. 4. Lo Bello M, Di Fini F, Notaro A, et al. ALS-related mutant FUS protein is mislocalized to cytoplasm and is recruited into stress granules of fibroblasts from asymptomatic Disclosure FUS P525L mutation carriers. Neurodegener Dis 2017;17:292–303. 5. Chi`o A, Restagno G, Brunetti M, et al. Two Italian kindreds with familial amyotrophic Disclosures available: Neurology.org/NG. lateral sclerosis due to FUS mutation. Neurobiol Aging 2009;30:1272–1275. 6. Ishigaki S, Sobue G. Importance of functional loss of FUS in FTLD/ALS. Front Mol Biosci 2018;5:44. Publication history ’ fi 7. Par´e B, Gros-Louis F. Potential skin involvement in ALS: revisiting Charcot s Received by Neurology: Genetics November 7, 2019. Accepted in nal observation—a review of skin abnormalities in ALS. Rev Neurosci 2017;28: form January 31, 2020. 551–572.

Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 CLINICAL/SCIENTIFIC NOTES OPEN ACCESS Multisystem mitochondrial disease caused by a rare m.10038G>A mitochondrial tRNAGly (MT-TG) variant

Olivia V. Poole, MD,* Alejandro Horga, MD, PhD,* Steven A. Hardy, PhD, Enrico Bugiardini, MD, Correspondence Cathy E. Woodward, BSc, Iain P. Hargreaves, PhD, FRCPath, Ashirwad Merve, PhD, FRCPath, Dr. Pitceathly [email protected] Rosaline Quinlivan, FRCP, Robert W. Taylor, PhD, FRCPath, Michael G. Hanna, FRCP, and Robert D.S. Pitceathly, PhD, MRCP

Neurol Genet 2020;6:e413. doi:10.1212/NXG.0000000000000413

Most pathogenic mitochondrial DNA (mtDNA) variants occur in the 22 mtDNA-encoded tRNA (mt-tRNA) genes. However, despite more than 270 reported mt-tRNA gene mutations, only 5 reside within mt-tRNAGly (MT-TG).1 We report a rare MT-TG variant and evaluate this, in addition to all previously reported MT-TG variants, against the published criteria used to help determine the pathogenicity of the mt-tRNA variants.2

Case report A 39 year old woman, born to nonconsanguineous parents, was reviewed in a specialist mitochondrial disorders clinic. She presented with hearing loss in her late teens followed by visual impairment, with bilateral cataracts, retinal dystrophy, and subsequent bilateral retinal detachments in her twenties; hypothyroidism in her thirties; and secondary amenorrhea. Clinical examination was otherwise normal, apart from short stature. There was no family history of neuromuscular or neurologic disease. Blood tests, including creatine kinase, plasma amino acids, acylcarnitine profile, very long chain fatty acids, and white cell enzymes, were normal. Plasma lactate was elevated (3.70 mmol/L, reference range 0.5–2.2 mmol/L). Nerve conduction studies and EMG showed no evidence of generalized myopathy or large fiber neuropathy. Histochemical analyses of muscle tissue revealed ragged-red and cyto- chrome c oxidase (COX) deficient fibers (figure A). Spectrophotometric determination of mitochondrial respiratory chain enzyme activities as a ratio to citrate synthase activity3 confirmed decreased activities of complexes I (0.076, reference range 0.104–0.268) and IV (0.006, reference range 0.014–0.034). Analysis of the next generation sequencing data (Ilumina MiSeq) of the entire mitochondrial genome extracted from the muscle3 revealed a rare m.10038G>A variant (GenBank reference accession number: NC_012920.1) in MT- TG (figure B) that was present at variable heteroplasmy levels across tissue types: 15% blood, 40% urinary epithelial cells, and 92% skeletal muscle. Maternal transmission was confirmed: 3% mutant load was present in the mother’s urinary epithelial cells (method- ology detects heteroplasmy levels ≥1%). Heteroplasmy levels within individual laser- captured COX-positive and COX-deficient muscle fibers were quantified by pyrosequenc- ing.4 Single fiber segregation studies confirmed a higher mutation load in COX-deficient (mean 95.30% ± 0.50%, SD, n = 27) compared with COX-positive fibers (mean 78.92% ± 4.43%, SD, n = 26; p = 0.0005, figure C).

*These authors contributed equally to the manuscript. From the Department of Neuromuscular Diseases (O.V.P., A.H., E.B., R.Q., M.G.H., R.D.S.P.), UCL Queen Square Institute of Neurology and The National Hospital for Neurology and Neurosurgery, London, United Kingdom; Wellcome Centre for Mitochondrial Research (S.A.H., R.W.T.), Translational and Clinical Research Institute, Newcastle University, Newcastle Upon Tyne; Neurogenetics Unit (C.E.W.), and Neurometabolic Unit (I.P.H.), The National Hospital for Neurology and Neurosurgery; Division of Neuropathology (A.M.), UCL Queen Square Institute of Neurology; Department of Histopathology (A.M.), Camelia Botnar Laboratory, Great Ormond Street Hospital; and Dubowitz Neuromuscular Centre (R.Q.), Great Ormond Street Hospital, London, United Kingdom.

Go to Neurology.org/NG for full disclosures. Funding information is provided at the end of the article.

The Article Processing Charge was funded by the authors. This is an open access article distributed under the terms of the Creative Commons Attribution License 4.0 (CC BY), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Copyright © 2020 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Figure Characterization of a rare pathogenic MT-TG variant

(A) Staining of the muscle sections with haematoxylin and eosin demonstrated ragged-red fibers (*) which were confirmed in the Gomori trichrome preparation (*) with the inset showing high magnification of left top fiber. A number of fibers, often with ragged-red morphology, showed a mild increase in the number of lipid droplets with Sudan black (*). Sequential COX and SDH histochemistry demonstrated frequent COX-deficient fibers, some of which had a ragged red–like appearance (*). The bar represents 100 μm for all stains with the inset 50 μm. (B) Two-dimensional cloverleaf structure of mitochondrial DNA-encoded tRNAGly. Black arrows indicate previously reported pathogenic variants. Red arrow indicates the rare variant identified in our patient. (C) Single muscle fiber segregation studies. COX-deficient fibers, (blue circles) harbor a higher mutational load compared with COX-positive fibers (red circles). (D) Evolutionary conservation of the m.10038G residue across species. COX = cytochrome c oxidase; SDH = succinate dehydrogenase.

Discussion that only 2 variants; m.9997T>C and m.10010T>C, should be considered “definitely pathogenic.” Although the m.9997T>C The m.10038G nucleotide is highly conserved across species variant has only been reported in one family, trans- (figure D, figure e-1, links.lww.com/NXG/A248) and the G>A mitochondrial cybrid studies support its pathogenic effects. transition at this location disrupts a C-G Watson-Crick base The m.10010T>C variant has been reported a number of pair in the TΨC stem of the tRNA molecule. The times, and its pathogenicity has been confirmed using single m.10038G>A variant is rare (absent from 3,450 in-house and fiber segregation studies. The m.10006A>G variant has only 1 48,882 GenBank sequences) and detectable at variable het- been reported in individuals harboring additional mtDNA eroplasmy levels across different tissues, with the highest levels variants, and thus may represent a benign polymorphism or be in the postmitotic muscle. Histochemical and biochemical ev- insufficient to cause disease in isolation. Current evidence idence of reduced complex I and IV activities in the muscle is suggests that the m.10014A>G variant is a benign poly- supportive of impaired mitochondrial protein synthesis, morphism. Finally, although the scoring system indicates that whereas single fiber studies confirmed segregation of the COX the m.10044A>G variant is “possibly pathogenic,” it has been defect with higher mutant levels. Consequently, the mutation detected in healthy controls, and thus may represent a hap- would be considered “definitely pathogenic” based on the ac- logroup specific polymorphism.5 cepted criteria for assigning pathogenicity to tRNA mutations (table e-1, links.lww.com/NXG/A249).2 It remains unclear why pathogenic variants occur more fre- quently in specific mt-tRNA genes. Possible explanations in- Review of the 5 previously reported MT-TG variants according clude intensive investigation of mt-tRNAs in which pathogenic to these criteria (table e-1, links.lww.com/NXG/A249) reveals variants have previously been identified, and survivor bias and

2 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG the absence of maternal transmission of mutations in mt-tRNA Disclosure genes linked with severe biochemical deficiencies, with isolated The present study is not industry sponsored. O.V. Poole, cases dissuading clinicians from further investigation of an A. Horga, S.A. Hardy, E. Bugiardini, C. Woodward, I.P. underlying mitochondrial disorder.6 Variants in mt-tRNAGly Hargreaves, A. Merve, R. Quinlivan, R.W. Taylor, M.G. affecting its canonical or noncanonical functions may also po- Hanna, and R.D.S. Pitceathly report no disclosures. Go to tentially be better tolerated than other mt-tRNAs or, con- Neurology.org/NG for full disclosures. versely, be severely deleterious and embryonic lethal. Publication history The detection of the rare mt-tRNA gene variants in suspected Received by Neurology: Genetics July 7, 2019. Accepted in final form mitochondrial disease has become more commonplace, given January 23, 2020. the widespread availability of whole mtDNA high throughput sequencing. Furthermore, whole genome sequencing, which Appendix Authors includes capture and deep sequencing of the mitochondrial genome, is identifying mtDNA variants in those in whom mi- Name Location Contribution tochondrial disease may not previously have been considered. Olivia V. UCL Queen Square Study concept and design, However, despite advances in the DNA sequencing technol- Poole, MD Institute of Neurology and Major role in acquisition of The National Hospital for data, Analysis/ ogy, the challenge of assigning pathogenicity to the rare Neurology and interpretation of data, mtDNA variants remains. It is therefore crucial that pathogenic Neurosurgery, London, Revising the manuscript variants are reported after confirmation using “gold-standard” United Kingdom for intellectual content techniques, for example, single fiber or transmitochondrial Alejandro UCL Queen Square Institute Study concept and design, Horga, MD of Neurology and The Major role in acquisition of cybrid studies, given the implications to genetic counseling and National Hospital for data, Analysis/ the available reproductive options for mtDNA mutations, in- Neurology and interpretation of data, Neurosurgery, London, Revising the manuscript cluding mitochondrial donation. Moreover, generating a com- United Kingdom for intellectual content prehensive data set for “definitely pathogenic” mt-tRNA variants would potentially advance the understanding of the Steven A. Wellcome Centre for Major role in acquisition of Hardy, PhD Mitochondrial Research, data, Analysis/ molecular mechanisms underpinning the susceptibility of in- Translational and Clinical interpretation of data, dividual genes to deleterious variants and facilitate the de- Research Institute, Newcastle Revising the manuscript University, Newcastle Upon for intellectual content velopment of targeted therapies to treat this group of disorders. Tyne, United Kingdom

Enrico UCL Queen Square Institute Major role in acquisition of Acknowledgment Bugiardini, of Neurology and The data, Analysis/ The authors would like to sincerely thank the family that MD National Hospital for interpretation of data, Neurology and Drafting/revising the participated in the study. Neurosurgery, London, manuscript for intellectual United Kingdom content Study funding Cathy E. The National Hospital for Major role in acquisition of Part of this work was undertaken in the University College Woodward, Neurology and data, Analysis/ London Hospitals/University College London Queen Square BSc Neurosurgery, London, interpretation of data, United Kingdom Revising the manuscript Institute of Neurology sequencing facility, which received for intellectual content a proportion of funding from the Department of Health’s Iain P The National Hospital for Major role in acquisition of National Institute for Health Research Biomedical Research Hargreaves, Neurology and data, Analysis/ Centres funding scheme. This research was supported by the PhD, Neurosurgery, London, interpretation of data, National Institute for Health Research University College FRCPath United Kingdom Drafting/revising the manuscript for intellectual London Hospitals Biomedical Research Centre. O.V. Poole content has received funding from the Lily Foundation. R.W. Taylor is Ashirwad UCL Queen Square Major role in acquisition of supported by the Wellcome Centre for Mitochondrial Re- Merve, PhD, Institute of Neurology, data, Analysis/interpretation search (203105/Z/16/Z), the Medical Research Council FRCPath London, United Kingdom of data, Drafting/revising the manuscript for intellectual (MRC) International Centre for Genomic Medicine in content Neuromuscular Disease, Mitochondrial Disease Patient Co- Rosaline UCL Queen Square Analysis/interpretation of hort (UK) (G0800674), the UK NIHR Biomedical Research Quinlivan, Institute of Neurology and data, Drafting/revising the Centre for Ageing and Age-related disease award to the FRCP The National Hospital for manuscript for intellectual Neurology and content Newcastle upon Tyne Foundation Hospitals NHS Trust, the Neurosurgery, London, MRC/EPSRC Molecular Pathology Node and The Lily United Kingdom

Foundation. R.D.S. Pitceathly is supported by a Medical Re- Robert W. Wellcome Centre for Major role in acquisition of search Council Clinician Scientist Fellowship (MR/S002065/ Taylor, PhD, Mitochondrial Research, data, Analysis/ 1). The clinical and diagnostic mitochondrial services in FRCPath Translational and Clinical interpretation of data, Research Institute, Drafting/revising the London and Newcastle upon Tyne are funded by the UK Newcastle University, manuscript for intellectual NHS Highly Specialised Commissioners to provide the “Rare Newcastle Upon Tyne, content United Kingdom Mitochondrial Disorders” Service. Continued Neurology.org/NG Neurology: Genetics | Volume 6, Number 2 | April 2020 3 References Appendix (continued) 1. Lott MT, Leipzig JN, Derbeneva O, et al. mtDNA variation and analysis using mitomap and mitomaster. Curr Protoc Bioinformatics 2013;44:1–23. Name Location Contribution 2. Yarham JW, Al-Dosary M, Blakely EL, et al. A comparative analysis approach to de- termining the pathogenicity of mitochondrial tRNA mutations. Hum Mutat 2011;32: 1319–1325. Michael G. UCL Queen Square Institute Study concept and design, Hanna, FRCP of Neurology and The Analysis/interpretation of 3. Poole OV, Everett CM, Gandhi S, et al Adult-onset Leigh syndrome linked to the novel stop codon mutation m. 6579G> A in MT-CO1. Mitochondrion 2019;47: National Hospital for data, Revising the – Neurology and Neurosurgery, manuscript for intellectual 294 297. London, United Kingdom content 4. Blakely EL, Yarham JW, Alston CL, et al. Pathogenic mitochondrial tRNA point mutations: nine novel mutations affirm their importance as a cause of mitochondrial – Robert D.S. UCL Queen Square Study concept and design, disease. Hum Mutat 2013;34:1260 1268. Pitceathly, Institute of Neurology and Analysis/interpretation of 5. Lehtonen MS, Meinil¨a M, Hassinen IE, Majamaa KJ. Haplotype-matched controls as PhD The National Hospital for data, Revising the a tool to discriminate polymorphisms from pathogenic mutations in mtDNA. Hum – Neurology and manuscript for intellectual Genet 1999;105:513 514. Neurosurgery, London, content 6. McFarland R, Elson JL, Taylor RW, Howell N, Turnbull DM. Assigning pathogenicity United Kingdom to mitochondrial tRNA mutations: when “definitely maybe”is not good enough. Trends Genet 2004;20:591–596.

4 Neurology: Genetics | Volume 6, Number 2 | April 2020 Neurology.org/NG CORRECTION Heritability of cervical spinal cord structure Neurol Genet 2020;6:e419. doi:10.1212/NXG.0000000000000419

In the article “Heritability of cervical spinal cord structure” by Dahlberg et al.,1 first published online February 26, 2020, figures 2 and 3 should have been switched to match their correct legends. The editorial office regrets the error.

Reference 1. Dahlberg LS, Viessmann O, Linnman C. Heritability of cervical spinal cord structure. Neurol Genet 2020;6:e401. doi: 10.1212/ NXG.0000000000000401.

Copyright © 2020 American Academy of Neurology 1 Copyright © 2020 American Academy of Neurology. Unauthorized reproduction of this article is prohibited.