Fewer Thymic Changes inMuSK Antibody-Positivethan in MuSK Antibody-Negative MGMaria Isabel Leite, MD,1 Philipp Strobel, MD,2

Margaret Jones, MSc,3 Kingsley Micklem, PhD,3

Regina Moritz,2 Ralf Gold, MD,2 Erik H. Niks, MD,4

Sonia Berrih-Aknin, PhD,5 Francesco Scaravilli, FRCPath,6

Aurea Canelhas, MD,7 Alexander Marx, PhD,2

John Newsom-Davis, FRS,1 Nick Willcox, PhD,1

and Angela Vincent, FRCPath1

In generalized myasthenia gravis (MG) patients withoutdetectable acetylcholine receptor (AChR) antibodies(SNMG), the thymus is often reported as “normally invo-luted.” We analyzed thymic compartments in 67 patientswith generalized MG, with AChR antibodies (AChR�,n � 23), with muscle-specific kinase (MuSK) antibodies(MuSK�, n � 14) or with neither (MuSK�, n � 30), andin 11 non-MG controls. Four of 14 MuSK� thymi hadrare small germinal centers, but overall they were not dif-ferent from age-matched controls. However, approximately75% MuSK� samples showed lymph node–type infiltratessimilar to those in AChR� patients, but with fewer ger-minal centers. These variations may explain some apparentdifferences in responses to thymectomy in SNMG.

Ann Neurol 2005;57:444–448

More than 80% of patients with typical generalizedmyasthenia gravis (MG) have detectable IgG autoanti-bodies against the native muscle acetylcholine receptor(AChR).1,2 Some AChR antibody seronegative(SNMG) patients have, instead, IgG autoantibodiesagainst the muscle-specific kinase (MuSK).3–6 Theirsymptoms tend to be more severe, with marked bulbar

involvement,4–7 and they often require high-dose im-munosuppressive treatment.4,5 The remaining SNMGpatients are negative for both autoantibodies.

The thymus is thought to play an important patho-genetic role in AChR antibody–positive (AChR�)younger patients, whose myasthenia often improves af-ter early thymectomy.8 In these cases, muscle-like my-oid cells in the thymic medulla are implicated in theperivascular infiltration by lymph node–type T-cell ar-eas and germinal centers.9–13 However, in SNMG, thethymus histology is frequently reported as “involuted”or “atrophic.” Although some abnormalities have beenreported previously in small series,9,14,15 their relation-ship with the recently reported MuSK autoantibodieshas not been explored. Here, we analyzed quantita-tively the thymic changes in a large series of SNMGpatients, stratified for MuSK antibody status, compar-ing with AChR� and age-matched control tissue.

Patients and MethodsClinical MaterialWe studied thymic tissue from 67 generalized MG patients,diagnosed by clinical and electromyographic criteria and un-dergoing thymectomy in the authors’ institutes (Table). Weincluded all SNMG cases with generalized MG with avail-able serum and thymus sections, plus 23 consecutiveAChR� cases. The 11 adult controls were undergoingmostly thyroid or parathyroid surgery in Wurzburg. We re-assayed all SNMG sera for AChR and MuSK antibodies us-ing commercial antigens (RSR, Cardiff, UK). Eight previ-ously negative sera proved low-positive (0.5–2nM) in thisAChR assay (and negative against MuSK). For clarity, theirresults have been excluded from this analysis.

ImmunohistochemistryThymic sections (5�m) from single routine formalin-fixed,paraffin-embedded blocks were dewaxed and rehydrated andmicrowaved in Target Retrieval Solution (DakoCytomation,Glostrup, Denmark) for 10 minutes at 900W. Slides wereincubated at 20°C in a peroxidase-blocking reagent for 10minutes (DakoCytomation), and for 30 minutes with anti-bodies to human CD3 (mature T cells, rabbit Ab; Dako-Cytomation), CD1a (immature thymocytes, mouse IgG1mAb, clone O10; Immunotech, Marseille, France), or cyto-keratin (epithelial cells, clones LP34 and MNF 116, bothmouse IgG1 from Dako-Cytomation). Binding was detectedwith the peroxidase-based Envision� (DakoCytomation)method; counterstaining was with hematoxylin.

Tissue MeasurementsImages were acquired by scanning each entire section (whichexcludes observer bias) on a transilluminated flatbed scanner(Eversmart Pro, Creo Scitex) at 124 pixels/mm, and imageswere analyzed using Photoshop (Adobe) with Image Process-ing Toolkit plug-ins (Reindeer Graphics, Asheville, NC). Wemeasured the areas of total tissue (including fat/connectivetissue), thymic tissue (defined as CD3�), thymic cortex(CD1a�), medulla (CD1a�/CD3�/keratin�), and infiltrates

From the 1Neurosciences Group, Department of Clinical Neurol-ogy, University of Oxford, United Kingdom; 2Departments of Pa-thology and Neurology, University of Wurzburg, Wurzburg , Ger-many; 3LRF Immunodiagnostics Unit, Nuffield Department ofClinical and Laboratory Sciences, University of Oxford, UnitedKingdom; 4Department of Neurology, Leiden University MedicalCenter, RC Leiden, The Netherlands; 5CNRS UMR 8078, UPS,IPSC, Hopital Marie Lannelongue, Le Plessis-Robinson, France;6Division of Neuropathology, National Hospital for Neurology andNeurosurgery, London, United Kindom; and 7Department of Pa-thology, Hospital Santo Antonio, Porto University, Porto, Portugal.

Received Sep 22, 2004, and in revised form Dec 3. Accepted forpublication Dec 8, 2004.

Current address for Dr Gold: Institute for MS Research, MedicalFaculty and Gemeinnutzige Hertie Stiftung, Waldweg 33, 37077Gottingen, Germany.

Published online Feb 24, 2005, in Wiley InterScience(www.interscience.wiley.com). DOI: 10.1002/ana.20386

Address correspondence to Dr Vincent, Neurosciences Group, De-partment of Clinical Neurology, University of Oxford, OX3 9DU,UK. E-mail: avincent@hammer.imm.ox.ac.uk

444 © 2005 American Neurological AssociationPublished by Wiley-Liss, Inc., through Wiley Subscription Services

(CD1a�/CD3�/keratin�), after converting the image to abinary form by thresholding.

Double Immunofluorescence Labeling ofGerminal CentersMicrowaved paraffin sections were incubated for 30 minutesat 20°C with a mixture of mouse mAbs to CD20 (B cells;clone L26, IgG2a) and CD35 (CR1 complement receptor ongerminal center B cells and follicular dendritic cells; cloneBer-Mac-DRC, IgG1, both from DakoCytomation) andthen with isotype-specific secondary antibodies conjugated toAlexa Fluor 488 or 568 (Molecular Probes, Leiden, TheNetherlands). Slides were mounted, and nuclei were stainedwith 4,6-diamidino-2-phenylindole, in fluorescence mount-ing medium (DakoCytomation). Germinal centers werecounted over entire sections, and their frequencies/mm2 ofthymic tissue were calculated. Images were captured via acooled digital camera. We used paraffin sections of tonsils aspositive controls, and omission of primary antibodies as neg-ative controls.

Statistical AnalysisWe used the Kruskal–Wallis one-way analysis of variancetest, followed by Dunn’s post-test, linear regression, and �2

with Yates’ correction.

ResultsClinical Features of the Myasthenia Gravis PatientsThe AChR�, MuSK�, or MuSK� MG subgroupsshowed no significant differences in ages at MG onsetor thymectomy, or in MG duration, although agestended to be lower in MuSK� patients (see Table).

Thymic HistologyExamples of thymus histology are shown in Figure 1and the morphometric data are summarized graphicallyin Figure 2. The most striking findings were the sim-ilarities between thymi from AChR� patients andmany MuSK� patients, and between the majority ofMuSK� and control samples (see Fig 1). To analyzethe differences, we assessed thymic tissue as a percent-age of total tissue, and thymic cortex as a percentage of

thymic tissue. In the controls, thymic tissue constitutedfrom 8 to 45%, decreasing with age as expected (r ��0.8: p � 0.01), and cortex ranged from 25 to 72%of the thymic tissue. As expected from previous studies,there was a significantly higher proportion of thymictissue in the AChR� than control samples (p � 0.05;see Fig 2A), and a lower percentage of cortex (p �0.001; see Fig 2B). In the MuSK� samples, both setsof values spanned a wide range, some overlapping theAChR� and others the control ranges (see Fig 2A, B).In contrast, the MuSK� fell within the control rangesfor both measurements (see Fig 2A, B), sometimes alsoshowing premature fatty replacement (see Fig 1D).

The areas of perivascular lymph node–type infiltrateswere increased in the AChR� compared with normalsamples, as expected,9–12,16 but also in approximately75% of the MuSK� patients (see Figs 1B and 2C). Incontrast, infiltrates were within normal limits in 10 of14 MuSK� thymi (eg, see Fig 1D), despite their rel-atively early ages at thymectomy. Germinal centerswere prominent in the infiltrates in the AChR� thymi(see Figs 1B, F and 2D), and they were variable butless frequent in the MuSK� thymi, whose infiltratesoften consisted mainly of T-cell areas (cf Fig 2C, D).Germinal centers were found in only 4 of 14 MuSK�thymi (see Fig 2D) and then were small (see Fig 1H).When all MG and control results were combined, thepercentage of infiltrates correlated positively with ger-minal center frequencies (r � 0.6; p � 10�8) and neg-atively with the percentage of cortex (r � �0.5; p �10�5).

Although overall there was a significant reduction ofthymic tissue in the 20 patients receiving steroid ther-apy compared with the 47 untreated patients (p �0.05), the differences between the MG subgroups re-mained (not shown).

DiscussionOnly mild abnormalities were noted in the three pre-vious studies on thymic pathology in SNMG.9,14,15

Table. Demographics of Patients and Controls

CharacteristicsControls(n � 11)

AChR AntibodyPositive

(n � 23)

MuSK AntibodyNegative(n � 30)

MuSK AntibodyPositive

(n � 14)

Sex ratio F:M 8:3 21:2 22:8 10:4MG onset-age, yr, median (range) 24.0 (12–44) 27.5 (10–47) 20.5 (10–43)Age at thymectomy, yr, median (range) 31.0 (17–40) 25.0 (14–44) 32.0 (10–54) 21.5 (12–44)MG duration at thymectomy, mo, median,

(range) 18.0 (4–94) 9.0 (1–204) 12.0 (2–72)Steroids at thymectomy, number 0 6/23 8/30 6/14

There was no significant difference between these groups for age at onset, age at thymectomy, duration of disease, or corticosteroid therapy(p � 0.5).

AChR � acetylcholine receptor; MuSK � muscle-specific kinase; MG � myasthenia gravis.

Leite et al: Thymus in Seronegative MG 445

Now, by stratifying a much larger series according toMuSK antibody status, we show that the thymus isnormal or only mildly altered in patients with MuSKantibodies, further emphasizing their pathogenetic aswell as clinical differences from typical MG. Strikingly,however, 75% of MuSK� patients had infiltrates andmany also had germinal centers, features typical of theAChR� MG thymus.

Only 14 (32%) of our 44 SNMG patients, from sixcenters in Northern Europe, were MuSK antibody–positive, reflecting growing evidence of lower preva-lences in Northern than in Southern Europe4 (A. Vin-cent, unpublished data). Because human myoid cellsapparently express MuSK,17 one might have perivascu-lar infiltrates as found in AChR� thymi, but we foundonly small infiltrates and only in 4 of 14 MuSK� pa-tients. These infiltrates may in fact belong within therange of normality, because larger series have shownconsiderable variability among control thymi.16 Never-theless, these changes could reflect an attack on myoidcells by the IgG1 complement-fixing antibodies foundin some patients,6 or possibly by the IgM-like factorfound in some MuSK� as well as MuSK� cases.18

The MuSK� cases proved to be heterogeneous, butmany showed clear thymic abnormalities. Indeed, theirmarked infiltrates appear very similar to those in typi-cal AChR� MG, but because they include fewer ger-minal centers, T-cell areas appear more prominent, aspreviously noted in SNMG patients overall.14 Thatmay imply some differences in the autoimmune re-sponse in MuSK� MG. Nevertheless, if these changesalso prove to be centered on myoid cells as in AChR�MG,12 that might further implicate myoid cells in au-toimmunization.

We had excluded eight patients who previously hadbeen seronegative but proved weakly AChR antibody–positive with the current more sensitive assay. Interest-ingly, their thymi showed the same range of changes asin the persistently AChR antibody–negative/MuSK�patients, so they may belong to the same spectrum.Continued improvements to the AChR antibody assaymay lead to further reductions in the number of AChRantibody–negative patients.

There are no controlled trials of thymectomy inSNMG. Some centers (eg, Oxford) seldom performsurgery, partly because of early experience with patients

Fig 1. The different thymic compartments in myasthenia gravis (MG) patient subsets. (A–D) Immunoperoxidase labeling for cytokeratin(red), showing cortex (C) as densely packed areas at the periphery of each lobule, medulla (M) with more intensely stained epithelialcells, and infiltrates as keratin-negative areas (arrows) that often include germinal centers (arrowheads). Interlobular fat (unstained) issurprisingly abundant in the muscle-specific kinase (MuSK)� case (D), despite her youth. Bar � 500�m. (A) Control (female aged40 years); (B) acetylcholine receptor (AChR)� donor (female aged 17 years); (C) MuSK� donor (male aged 38 years); (D) MuSK�donor (female aged 15 years): none of these patients had received corticosteroids. (E–H) Double immunofluorescence labeling for CD20(green) and CD35 (red). In the control (E), the medullary B cells clustered around the Hassall’s corpuscles (HC) label only for CD20,whereas the follicular dendritic network in the germinal centers (arrowheads in F–H) is also CD35-positive. Germinal centers are typi-cally larger in AChR� MG (F) than in MuSK� and MuSK� cases (G, H); they were seen in only 4 of the 14 MuSK� samples(see Fig 2), one of which is shown in panel H. Bar � 100�m. (E) Control (female aged 40 years); (F) AChR�MG donor (femaleaged 20 years); (G) MuSK�donor (female aged 43 years); (H) MuSK� donor (female aged 19 years); none had receivedcorticosteroids.

446 Annals of Neurology Vol 57 No 3 March 2005

now found to have MuSK antibodies (A. Vincent andJ. Newson-Davis, unpublished observations) and al-most normal thymi. Conversely, the striking similari-ties to AChR� MG in many MuSK� thymi are con-sistent with the possible responses to thymectomy seenin some AChR antibody–negative patients. In futureanalyses, negativity for MuSK autoantibodies, and evenpreoperative assessment of thymic infiltrates (eg, withlabeled mAbs to CD21, CD35 and/or CD72),19 mighthelp to predict thymic histology and potential re-sponses to thymectomy.

This work was supported by Fundacao para a Ciencia e a Tecnolo-gia, Portugal (M.I.L.).

We are very grateful to the Myasthenia Gravis Association and theMuscular Dystrophy Campaign for their ongoing support. We arealso grateful to Drs A. R. Wintzen, J. J. G. M. Verschuuren, and A.Martins Silva for clinical data and samples, Dr S. G. van Duinenand L. Herbert for thymic sections, and J. B. Sousa for advice in thestatistical analysis.

References

1. Drachman DB. Myasthenia gravis. N Engl J Med 1994;330:1797–1810.

2. Vincent A. Unravelling the pathogenesis of myasthenia gravis.Nat Rev Immunol 2002;2:797–804.

3. Hoch W, McConville J, Helms S, et al. Auto-antibodies to thereceptor tyrosine kinase MuSK in patients with myasthenia gra-vis without acetylcholine receptor antibodies. Nat Med 2001;7:365–368.

4. Evoli A, Tonali PA, Padua L, et al. Clinical correlates with anti-MuSK antibodies in generalized seronegative myasthenia gravis.Brain 2003;126:2304–2311.

5. Sanders DB, El-Salem K, Massey JM, et al. Clinical aspects ofMuSK antibody positive seronegative MG. Neurology 2003;60:1978–1980.

6. McConville J, Farrugia ME, Beeson D, et al. Detection andcharacterization of MuSK antibodies in seronegative myastheniagravis. Ann Neurol 2004;55:580–584.

7. Zhou L, McConville J, Chaudhry V, et al. Clinical comparisonof muscle-specific tyrosine kinase (MuSK) antibody-positiveand -negative myasthenic patients. Muscle Nerve 2004;30:55–60.

Fig 2. Quantitation of thymic compartments in myasthenia gravis (MG) patient subsets. Panel A shows the percentage area of thymiccortex � medulla � infiltrates per total (including fat and connective tissue). The percentage area of thymic tissue occupied by cortex isshown in B and by infiltrates in C; D shows the number of germinal centers/mm2 of thymic tissue. The bars are the medians. All sig-nificant differences are shown in the figure. * p � 0.05; **p � 0.01; ***p � 0.001.

Leite et al: Thymus in Seronegative MG 447

8. Gronseth GS, Barohn RJ. Practice parameter: thymectomy forautoimmune myasthenia gravis (an evidence-based review): re-port of the Quality Standards Subcommittee of the AmericanAcademy of Neurology. Neurology 2000;55:7–15.

9. Berrih S, Morel E, Gaud C, et al. Anti-AChR antibodies, thy-mic histology, and T cell subsets in myasthenia gravis. Neurol-ogy 1984;34:66–71.

10. Schluep M, Willcox N, Ritter MA, et al. Myasthenia gravisthymus: clinical, histological and culture correlations. J Auto-immunity 1988;1:445–467.

11. Muller-Hermelink HK, Marx A. Pathological aspects of malig-nant and benign thymic disorders. Ann Med 1999;31:5–14.

12. Roxanis I, Micklem K, McConville J, et al. Thymic myoid cellsand germinal center formation in myasthenia gravis; possibleroles in pathogenesis. J Neuroimmunol 2002:125:185–197.

13. Wekerle H, Ketelsen UP. Intrathymic pathogenesis and dualgenetic control of myasthenia gravis. Lancet 1977;1:678–680.

14. Willcox N, Schluep M, Ritter MA, Newsom-Davis J. The thy-mus in seronegative myasthenia gravis patients. J Neurol 1991;238:256–261.

15. Verma PK, Oger JJ. Seronegative generalized myasthenia gravis:low frequency of thymic pathology. Neurology 1992;42:586–589.

16. Flores KG, Li J, Sempowski GD, et al. Analysis of the thymicperivascular space during aging. J Clin Invest 1999;104:1031–1039.

17. Mesnard-Rouiller L, Bismuth J, Wakkach A, et al. Thymic my-oid cells express high levels of muscle genes. J Neuroimmunol2004;148:97–105.

18. Plested CP, Tang T, Spreadbury I, et al. AChR phosphoryla-tion and indirect inhibition of AChR function in seronegativeMG. Neurology 2002;59:1682–1688.

19. Flores KG, Li J, Hale LP. B cells in epithelial and perivascularcompartments of human adult thymus. Hum Pathol 2001;32:926–934.

Subthalamic StimulationActivates Internal Pallidus:Evidence from cGMPMicrodialysis in PD PatientsAlessandro Stefani, MD,1,2 Ernesto Fedele, PhD,3,4

Salvatore Galati, MD,2 Olimpia Pepicelli, PhD,3

Stefania Frasca, MD,2

Mariangela Pierantozzi, MD, PhD,1,2

Antonella Peppe, MD, PhD,1 Livia Brusa, MD, PhD,2

Antonio Orlacchio, MD, PhD,1,2

Atticus H. Hainsworth, PhD,5 Giuseppe Gattoni, BSc,2

Paolo Stanzione, MD,1,2 Giorgio Bernardi, MD,1,2

Maurizio Raiteri, MD, PhD,3,4 and Paolo Mazzone, MD6

Parkinson’s disease patients benefit from deep brainstimulation (DBS) in subthalamic nucleus (STN), but thebasis for this effect is still disputed. In this intraoperativemicrodialysis study, we found elevated cGMP extracellu-lar concentrations in the internal segment of the globuspallidus, despite negligible changes in glutamate levels,during a clinically effective STN-DBS. This supports theview that a clinically beneficial effect of STN-DBS is par-alleled by an augmentation (and not an inactivation) ofthe STN output onto the GPi.

Ann Neurol 2005;57:448–452

Although subthalamic nucleus (STN) deep brain stim-ulation (DBS) has been accepted as a powerful clinicaloption in advanced Parkinson’s disease (PD) patients,we lack a precise understanding of the mechanisms ex-plaining DBS efficacy. Initially, it was inferred thatDBS acts through the inactivation of STN cells due tohigh-frequency stimulation (HFS). This view was sup-ported by animal1 and human studies.2 Several find-ings, however, have challenged this view. First, theability of human STN cells to fire at a peculiarly high

From 1Instituto di Ricovero e Cura a Carrattere Scientificio(IRCCS) Fondazione S. Lucia; 2Clinica Neurologica, Universita TorVergata; 3Dipartmento Medicina Sperimentale, Sezione di Farmaco-logia e Tossicologia, Universita degli Studi di Genova; 4Centro diEccellenza per la Ricerca Biomedica, Universita degli Studi diGenova, Genova, Italy; 5Pharmacology Research Group, the Leices-ter School of Pharmacy De Montfort University, Leicester, UnitedKingdom; and 6Divisione di Neurochirurgia, Ospedale CTO,Rome, Italy.

Received Aug 12, 2004, and in revised form Dec 17. Accepted forpublication Dec 20, 2004.

Published online Feb 24, 2005, in Wiley InterScience(www.interscience.wiley.com). DOI: 10.1002/ana.20402

Address correspondence to Prof. Stanzione, I.R.C.C.S. FondazioneS. Lucia, Via Ardeatina 306. 00179, Rome, Italy.E-mail: stanzione@med.uniroma2.it

448 © 2005 American Neurological AssociationPublished by Wiley-Liss, Inc., through Wiley Subscription Services