Brain (2001), 124, 2427–2438

Diagnostic value of sural nerve demyelination inchronic inflammatory demyelinatingpolyneuropathyW. M. J. Bosboom,1 L. H. van den Berg,1 H. Franssen,2 P. C. L. M. Giesbergen,1 H. Z. Flach,1

A. M. van Putten,1 H. Veldman1 and J. H. J. Wokke1

Departments of 1Neurology and 2Clinical Neurophysiology Correspondence to: Leonard H. van den Berg, MD,of the Rudolf Magnus Institute for Neurosciences, Department of Neurology, University Medical CenterUniversity Medical Center Utrecht, The Netherlands Utrecht, PO Box 85500, 3508 GA Utrecht, The Netherlands

E-mail: l.h.vandenberg@neuro.azu.nl

SummaryThe objective of the study was to determine the diagnosticvalue of features of demyelination and inflammation insural nerve biopsy specimens of patients with chronicinflammatory demyelinating polyneuropathy (CIDP). Thefeatures of (i) demyelination, (ii) axonal de- andregeneration and (iii) inflammation were investigated bymeasuring the number of onion bulbs, g ratio (axondiameter/total nerve fibre diameter), myelinated nervefibre density, number of clusters and endoneurial area in21 patients with CIDP, as well as in 13 patients withchronic idiopathic axonal polyneuropathy (CIAP) and sixautopsy controls. In addition, teased fibres were classifiedand lengths of internodes measured. We found nodifference in demyelinating features between patients withCIDP and CIAP, as the percentage of fibres with segmentalde- and remyelination and the number of onion bulbs

Keywords: chronic inflammatory demyelinating polyneuropathy; sural nerve; demyelination; axonal degeneration;pathology

Abbreviations: CIAP � chronic idiopathic axonal polyneuropathy; CIDP � chronic inflammatory demyelinatingpolyneuropathy

IntroductionChronic inflammatory demyelinating polyneuropathy (CIDP)is a progressive or relapsing disorder of the peripheral nervoussystem (Prineas and McLeod, 1976; McCombe et al., 1987;Barohn et al., 1989), associated with demyelination ofperipheral nerves and spinal roots (Bouchard et al., 1999).Diagnostic criteria for CIDP are based mainly on clinicaland electrophysiological features, but a sural nerve biopsy isoften taken to establish the diagnosis of CIDP (Cornblathet al., 1991). Segmental demyelination and remyelination,frequently resulting in onion bulb formation, together withinflammatory infiltrates is considered to be the pathologicalhallmark of CIDP (Krendel et al., 1989; Dyck et al., 1993a;

© Oxford University Press 2001

were similar in both polyneuropathy groups. The g ratio,expected to be higher in a demyelinating disease due tothinner myelin sheaths, was significantly lower in CIDPthan CIAP. Evidence for axonal degeneration was foundin both CIDP and CIAP, as both showed a decrease inmyelinated nerve fibre density. There was no evidence ofendoneurial oedema in CIDP, as the endoneurial area didnot differ between CIDP, CIAP and the autopsycontrols. Although significant differences of features ofdemyelination, axonal degeneration and inflammationwere found in sural nerve biopsy specimens, there was aconsiderable overlap between abnormalities in CIDP andCIAP patients. In the majority of patients, quantitativeanalysis of light microscopical abnormalities in suralnerves was similar in CIDP and CIAP. Therefore, a suralnerve biopsy is of limited diagnostic value in CIDP.

Uncini et al., 1999). De- and remyelinating features inCIDP are often accompanied by varying degrees of axonaldegeneration (Dyck et al., 1993a; Nagamatsu et al., 1999).Myelinated fibre loss, and evidence of sprouting, as well asthe presence of myelin ovoid formation in teased fibres, areconsidered as evidence of axonal degeneration (Krendel et al.,1989; Dyck et al., 1993b).

In many pathological studies, sural nerve biopsy specimensfrom patients with CIDP were not compared with well-defined control groups (Prineas and McLeod, 1976; Barohnet al., 1989; Matsumuro et al., 1994; Bouchard et al., 1999).We recently demonstrated that T cells were present in all

2428 W. M. J. Bosboom et al.

sural nerve biopsy specimens from patients with CIDP, butalso in non-inflammatory and autopsy controls. Moreover, incomparison with controls, most patients with CIDP did notdiffer with regard to numbers and distribution of T cells(Bosboom et al., 1999). The purpose of the present study wasto determine the diagnostic value of the classical pathologicalfeatures of de- and remyelination in CIDP.

Patients and methodsPatients and controlsWe investigated sural nerve biopsy specimens taken between1987 and 1995, from 21 consecutive patients who fulfilledestablished criteria for CIDP (Cornblath et al., 1991). Asdisease controls, we used sural nerves from 13 patients witha non-inflammatory chronic idiopathic axonal polyneuropathy(CIAP). CIAP is a slowly progressive polyneuropathy forwhich no cause could be found during a 5-year follow-up(Notermans et al., 1993, 1994, 1996; Teunissen et al., 1997).Patients with CIDP and CIAP were included in this studywithout using the biopsy findings to assign diagnoses. Theclinical, laboratory and electrophysiological findings of thepatients are summarized in Tables 1 and 2. As normalcontrols, we used sural nerves from six autopsy patientswithout known peripheral nerve disease, obtained within 24 hof death. Six of 21 patients with CIDP had been treatedbefore the biopsy was taken: Patients 3 and 4 had high-doseintravenous immunoglobulins 2 months previously, Patients2, 7 and 11 had prednisone �6 months previously, andPatient 1 had prednisone 1 month previously. The severityof the clinical picture was graded according to the modifiedRankin scale (van Swieten et al., 1988): 0 � normal;1 � signs but not symptoms of the neuropathy; 2 � mildmotor or sensory symptoms (or both) with or without mildfunctional impairment; 3 � moderately disabled by motorand sensory symptoms including ataxia; 4 � requiringassistance in eating or dressing, or using a walking aid; and5 � not ambulatory.

Sural nerve biopsy specimensSural nerves were obtained under local anaesthesia. The suralnerve was biopsied at the ankle above the lateral malleolus.Part of the nerve tissue was fixed in 2% bufferedglutaraldehyde, post-fixed in 1% buffered osmium tetroxideand dehydrated in acetone. The material was impregnatedwith and embedded in Epon and polymerized at 60°C for 24h. Transverse whole sural nerve sections of 1 µm were stainedwith alkaline toluidine blue or 1% p-phenylenediamine. Forteased fibre preparations, nerve tissue was fixed in 2%buffered glutaraldehyde, post-fixed in 1% buffered osmiumtetroxide, dehydrated in acetone and impregnated with Epon,and the nerve fibres were teased in a film of Epon on amicroscope slide.

Light microscopic analysisTransverse sections of the sural nerve biopsy specimens ofall patients with CIDP and CIAP, and autopsy controls were

examined to investigate features of (i) de- and remyelination,(ii) axonal de- and regeneration and (iii) inflammation. Teasedfibre preparations were available from 14 patients with CIDP(nos 1–14) and nine patients with CIAP (nos 22–30). Allinvestigations were performed in a blinded fashion.

Demyelinating featuresSegmental de- and remyelination. We performed a qualitativeanalysis of teased fibres in three different ways. First,segmental de- or remyelination was scored in all teasedfibres, ignoring other demyelinating or axonal features (Fig.1). The percentage of fibres with segmental demyelinationwas calculated and in these fibres the percentage of de- orremyelinated segments was determined. Secondly, the teasedfibres were classified according to the classification of Dyckand colleagues listed in Table 3 (Dyck et al., 1993b). Teasedfibres in category B were considered to be in an early stageof nerve damage not differentiating between a demyelinatingor axonal cause. Teased fibres in categories C, D, F and Gwere considered to be affected by a demyelinating process,and teased fibres in categories E and H were considered tobe affected by an axonal process. Category I was consideredto appear in Wallerian degeneration. Thirdly, two experiencedneuropathologists (F.G.I. Jennenens and G.H. Jansen) wereasked to examine all teased fibre preparations in a blindedfashion and to determine whether a demyelinating or axonalcause of the neuropathy was most likely.Onion bulbs. Onion bulbs were quantified in the totalendoneurial area by two independent observers (H.Z.F. andA.M.v.P.) using light microscopy (�40 objective) oftransverse sections stained with alkaline toluidine blue (Fig.1). An onion bulb was defined as a myelinated nerve fibreaxon surrounded by one or more concentrically orientatedsupernumerary Schwann cell processes (Low et al., 1978)for 100% of the circumference. The correlation coefficientfor the measurements of both observers of onion bulbs was0.86 (P � 0.01). For further analysis, the mean values of themeasurements of both observers were used.g ratio and diameter histograms. For each myelinated nervefibre, Feret-diameters (diameter of circle with the same areaas measured area) of the axon and of the total nerve fibrewere calculated. The g ratio was calculated (axon diameter/total nerve fibre diameter) and histograms of the totalmyelinated nerve fibre diameters and axon diameters weremade.

Axonal de- and regenerationMyelinated nerve fibres. Using a Zeiss Videoplan, myelinatednerve fibres were counted in 8–12 micrographs of randomlychosen areas of 1 µm transverse sections stained with alkalinetoluidine blue. These micrographs (magnification �2240)covered a total endoneurial area of between 0.06 and 0.09mm2 per biopsy. The myelinated nerve fibre density and totalnumber of myelinated nerve fibres were calculated.

Sural nerve demyelination in CIDP 2429

Table 1 Clinical data of patients with CIDP and CIAP

Patient Sex Age Nadir Biopsy Rankin CSF protein(years) (months) (months) (mg/dl)

Max Min

CIDP1 F 59 3 4 3 0 2802 F 35 8 15 2 2 2203 F 29 5 5 4 2 3304 M 53 22 28 4 1 1005 F 63 2 2 5 3 10306 F 49 5 5 5 4 3507 M 43 10 37 3 2 508 M 69 3 5 4 3 1309 M 76 8 3 5 4 160

10 M 48 3 3 3 2 10011 F 27 77 75 3 3 5012 M 59 15 15 2 1 13013 F 57 8 15 4 3 19014 M 24 6 4 4 1 10015 M 29 168 168 4 2 8016 F 54 2 2 4 1 13017 M 34 228 228 2 2 6018 M 53 4 11 2 1 20019 M 66 5 6 4 3 76020 M 55 18 30 3 1 16021 M 46 2 2 4 n.a. 140

Median 53 6 6 4 2 140CIAP

22 M 45 32 1 4623 F 67 12 3 n.a.24 M 55 22 3 3025 F 56 26 3 4626 M 48 11 2 n.a.27 M 60 97 3 5828 M 62 48 2 5629 F 68 26 2 6830 F 65 14 3 3231 M 54 29 1 n.a.32 M 61 19 3 7533 M 59 72 3 n.a.34 M 53 180 2 n.a.

Median 59 26 3 51P n.s. n.s. 0.02 �0.01 �0.01

CIDP � chronic inflammatory demyelinating polyneuropathy; CIAP � chronic idiopathic axonalpolyneuropathy; F � female; M � male; Biopsy � time from disease onset until biopsy;Rankin � modified Rankin scale; Max � maximal disability score; Min � minimal disabilityscore; P � P value of Mann–Whitney U-test used to compare CIDP with CIAP patients;n.a. � not available; n.s. � not significant.

Clusters. Clusters, considered as evidence of sprouting, werequantified in the total endoneurial area by two inde-pendent observers (H.Z.F. and A.M.v.P.) using lightmicroscopy (�40 objective) of transverse sections stained withp-phenylenediamine. A regenerative cluster was defined asthree or more closely apposed myelinated nerve fibres(Llewelyn et al., 1991), which had maximal variation in myelinthickness of 50%, a maximal distance between different fibresof 2.5 times the myelin thickness of the fibres and a myelinthickness of �50% of the thickest fibre in the fascicle. Thecorrelation coefficient for the measurements of both observers

was 0.96 (P � 0.01). For further analysis, the mean values ofthe measurements of both observers were used.

Internode length. In the teased fibre preparations, eachinternode length of all fibres was measured (objective �10)and the mean internode length of all fibres calculated.

Inflammatory featuresThe endoneurial area. As an increase in the endoneurial areamay be an indication of endoneurial oedema, the total

2430 W. M. J. Bosboom et al.

Table 2 Electrophysiological data of patients with CIDP and CIAP

Patient DML Conduct. vel. Amp. red. F-wave Distal CMAP amp Amp Spont. act.sur

Med Uln Med Uln Sur Med Uln Med Uln Med Uln Per Tib ID 1 TA

CIDP1 6.2 5.6 50 22 39 17 irr 45 38 11.2 0.9 2.6 n.a. 6 – –2 4.7 3.3 29 37 41 29 32 51 45 12.2 11.6 2.9 5.0 9 – –3 11.3 7.8 41 32 n.r. 27 5 n.r. 54 5.2 2.2 1.5 n.a. 0 – –4 4.9 3.3 36 39 50 14 27 39 47 11.1 7.7 4.0 4.1 12 – –5 n.a. 4.5 n.a. 60 n.r. n.a. 15 n.a. 36 n.a. 3.9 0.9 0.4 0 – �6 7.3 6.6 20 12 n.r. 36 irr 79 n.r. 3.3 0.1 0 1.4 0 – �7 5.8 5.6 28 22 n.r. 12 34 49 51 10.7 5.0 n.a. 0.3 0 – –8 n.a. 6.3 n.a. 26 n.r. n.a. 16 n.a. n.a. n.a. 6.8 n.a. 1.0 0 – �9 8.4 n.a. 21 n.a. n.r. 23 n.a. 78 n.a. 8.3 n.a. n.a. 0.5 0 – �

10 15.7 8.0 29 24 n.r. 18 8 n.r. 44 4.0 2.4 2.4 n.a. 0 – �11 6.9 6.0 28 31 51 irr 35 n.a. 46 0.4 2.7 0.3 n.a. 8 n.a. �12 7.8 5.9 43 40 n.r. –2 18 42 43 4.7 7.8 0.4 n.a. 0 – –13 5.5 5.0 30 23 n.r. 25 3 n.r. n.r. 4.0 3.0 n.a. 0 0 – �14 4.8 4.1 25 37 47 84 84 n.a. n.a. 9.5 8.8 n.a. 15.3 16 – –15 4.1 2.5 52 55 39 1 18 42 58 7.8 11.4 0.4 0.1 9 – �16 19.1 5.4 42 48 n.r. 12 15 49 36 3.3 8.1 1.2 n.a. 0 n.a. n.a.17 3.2 3.0 45 32 50 65 69 31 41 4.0 5.8 4.1 n.a. 10 – –18 7.1 5.5 50 50 45 0 11 38 37 6.0 5.7 0.3 n.a. 10 – �19 5.2 4.4 22 32 18 60 28 43 62 18.4 11.7 n.a. 1.3 8 � –20 13.2 13.6 22 17 n.r. 38 50 67 81 3.2 4.2 0 0 0 n.a. �21 16.3 4.9 57 52 38 0 0 46 36 2.7 4.2 0.6 1.3 6 – –

Median 6.9 5.5 30 32 43 21 18 46 45 5.2 5.4 0.9 1.0 0CIAP

22 n.a. 3.2 n.a. 49 36 n.a. 26 n.a. 36 n.a. 10.9 n.a. 0.3 6 � –23 n.a. n.a. 48 n.a. n.a. n.a. n.a. n.a. n.a. n.a. n.a. 0 n.a. n.a. n.a. �24 3.8 n.a. 53 n.a. 14 0 n.a. n.a. n.a. 10.0 n.a. n.a. 4. 1 1 – –25 n.a. 3.4 n.a. 55 n.r. n.a. 5 n.a. n.a. n.a. 3.9 n.a. 0.1 0 � �26 4.2 3.1 56 54 38 6 3 n.a. n.a. 12.2 9.4 5.7 6.4 5 n.a. –27 n.a. 3.2 n.a. 50 n.r. n.a. 13 n.a. 33 n.a. 7.6 n.a. 0 0 � �28 4.6 3.6 46 45 38 3 17 n.a. n.a. 7.6 10.7 n.a. 1.9 2 – –29 3.5 3.5 55 60 n.r. 0 24 n.a. n.a. 5.4 5.1 0.3 0.4 0 � �30 n.a. 3.2 n.a. 53 n.r. n.a. 16 n.a. 29 n.a. 6.3 n.a. 0.6 0 � �31 4.0 3.2 46 48 n.r. 9 11 34 33 16.5 8.8 0 1.9 0 – –32 n.a. 3.1 n.a. 51 n.r. n.a. 16 n.a. 31 n.a. 8.9 0 n.a. 0 � �33 4.5 3.3 52 54 n.r. 6 9 35 32 7.0 6.6 0.2 0.4 0 � �34 3.9 3.1 50 60 n.r. 4 9 34 32 9.9 9.0 0.1 2.6 0 – �

Median 4.0 3.2 52 52 37 4 13 34 32 9.9 8.8 0.1 0.6 0P 0.01 0.01 0.01 0.01 0.02 0.02 n.s. 0.02 0.01 n.s. n.s. n.s. n.s. n.s. 0.01 n.s.

Bold values are considered to be compatible with demyelination (Cornblath et al., 1991). DML � distal motor latency (ms);Med � median; Uln � ulnar, Per � peroneal, Tib � tibial, Sur � sural nerve; Conduct. vel. � conduction velocity (m/s); Amp. red. �amplitude reduction on elbow versus wrist stimulation (%); F-wave � shortest F-wave latency in the lower arm (ms); Distal CMAPamp � compound muscle action potential amplitude of negative peak on wrist or ankle stimulation (mV); Amp sur � sural nervesensory nerve action potential amplitude (mV); Spont. act. � presence (�) or absence (n) of spontaneous muscle fibre activity; ID 1 �interosseus dorsalis 1 muscle; TA � tibialis anterior muscle; P � P value of Mann–Whitney U-test used to compare CIDP with CIAPpatients; irr � irrelevant due to low distal CMAP; n.r. � no response; n.a. � not available; n.s. � not significant.

endoneurial area was measured in 1 µm transverse sectionsstained with alkaline toluidine blue, by light microscopy(�10 objective) using a digitizing tablet and image analysissoftware (Jandel). The sub-perineural area without myelinatednerve fibres, sometimes widened in CIDP biopsy specimens,was not included in the total endoneurial area.

T cells. Data on T cells from our previous study wereused, in which numbers and localization of CD3� cells intransverse sections of sural nerve biopsy specimens were

investigated. Patients 14 and 19 were not included in thisstudy (Bosboom et al., 1999).

Statistical analysisThe Mann–Whitney U-test was used to compare clinical,electrophysiological and pathological features betweenpatients with CIDP and control groups. This test was alsoused to compare the numbers of myelinated nerve fibres/axons with the same diameter between CIDP and the control

Sural nerve demyelination in CIDP 2431

Fig. 1 Segmental demyelination in teased fibres of Patients 5 (A) and 11 (B) with CIDP, and Patients23 (C) and 29 (D) with CIAP. Bar � 100 µm. Onion bulbs (arrows) in transverse sections of Patient 4with CIDP (E) and Patient 32 with CIAP (F). Bar � 10 µm

groups. Spearman’s rank correlation was used to calculatethe inter-observer variation of measurements on onion bulbsand clusters. A Fisher’s exact test was performed to determinewhether each neuropathologist was able to assign thediagnoses CIDP or CIAP based on qualification of the teasedfibres, and a Cohen’s kappa was calculated to measure theagreement between the two neuropathologists. To adjustdifferences in pathological findings between CIDP and CIAPfor age, time from onset of disease to moment of biopsy,

and endoneurial area, a logistic regression analysis wascarried out. In order to investigate which pathological featurehad the most significant influence on the differentiationbetween the diagnosis CIDP and CIAP, we entered allvariables with P � 0.20 into a multivariate model. Spearman’srank correlation was performed to correlate clinical (listedin Table 1) and pathological features within the group ofpatients with CIDP. P values �0.05 were considered to besignificant.

2432 W. M. J. Bosboom et al.

Fig. 2 The percentage of fibres with de- or remyelination (A), number of onion bulbs (B) and g ratio (C) in sural nerve biopsyspecimens of patients with CIDP and CIAP, and normal controls. Normal � autopsy controls; *,** � statistically different from patientswith CIDP (*P � 0.05, **P �0.01); long horizontal lines represent the upper and lower limits for CIAP and autopsy controls; the shorthorizontal line represents the median.

Table 3 Classification of teased fibres according to Dyck

(A) Teased fibres of normal appearance, ignoring criteria of internode length and of internode diameter. The average thickness ofmyelin of the internode with the thinnest myelin is 50% or more of that of the internode with the thickest myelin. No paranodalor internodal segmental demyelination is seen.

(B) Teased fibre with excessive irregularity, wrinkling, and folding of myelin but with the other features of condition A.(C) Teased fibre with a region or regions of paranodal or internodal segmental demyelination with or without ovoids or balls in the

cytoplasm of the associated Schwann cells. Thickness of myelin of the internode with the thinnest myelin is 50% or more of thatof the internode with the thickest myelin. Myelin of internodes may be regular or irregular.

(D) Teased fibre with a region or regions of paranodal or internodal segmental demyelination with or without myelin ovoids in thecytoplasm of the associated Schwann cells. Thickness of myelin of the internode with the thinnest myelin is less than 50% of thatof the internode with the thickest myelin. Myelin of internodes may be regular or irregular.

(E) Teased strand of nerve tissue with linear rows of myelin ovoids and balls at the same stage of degeneration.(F) Teased fibre without a region or regions of segmental demyelination but with excessive variability of myelin thickness among

internodes. Thickness of myelin of the internodes with the thinnest myelin is less than 50% of that of the internode with thethickest myelin.

(G) Teased fibre with excessive variability of myelin thickness within internodes to form ‘globules’or ‘sausages’.(H) Teased fibres of normal appearance as described in A, but in which there are myelin ovoids or balls contiguous to two or more

internodes.(I) Teased fibre having several proximal internodes or parts of internodes with or without paranodal or internodal segmental

demyelination and, distal to these, a linear row of myelin ovoids or balls.

ResultsFor all features of demyelination, axonal de- or regenerationand inflammation, we investigated whether the results inpatients with CIDP were significantly different from thosein patients with CIAP or autopsy controls (Figs 2–6).Subsequently we studied whether the values in individualCIDP patients were increased or decreased (i.e. higher orlower than the highest or lowest value in patients with CIAPor autopsy controls) (Table 4).

Demyelinating featuresSegmental de- and remyelinationThe length of the teased fibres in preparations of patientswith CIDP or CIAP was ~5 mm, and the number of fibresexamined ranged from 27 to 87 (median 47).

The percentage of teased fibres with segmental de- orremyelination and the number of de- or remyelinatingsegments per affected fibre did not differ significantly betweenCIDP and CIAP (Fig. 2A), which means that the distributionof the de- or remyelinated segments along the fibres wassimilar in CIDP and CIAP. Three patients with CIDP had ahigher, and one patient a lower percentage of fibres withsegmental de- or remyelination, than any of the CIAP controls(Fig. 2A and Table 4).

When the teased fibres were classified according to Dyckand colleagues (Table 3) (Dyck et al., 1993b), the percentageof normal fibres (category A) was lower in CIDP than CIAP(Fig. 3). Categories D and E were significantly more frequentin CIDP. When the percentages of categories C, D and F,which were all considered to be a feature of demyelination,

Sural nerve demyelination in CIDP 2433

Table 4 Comparison of demyelinating and inflammatorypathological features in patients with CIDP with patientswith CIAP and autopsy controls

CIPD Segmental Onion g ratio Area T cellspatient demyelination* bulbs

123 ↑ ↑4 ↑ ↑5 ↑ ↑ ↑6 ↓ ↑ ↑789 ↑

10 ↑11 ↑12 ↑13 ↑14 ↓ n.a.15 n.a. ↑16 n.a.17 n.a.18 n.a.19 n.a. n.a.20 n.a. ↑21 n.a. ↑

*Only compared with CIAP; Area � endoneurial area; T cells �number of T cells in total sural nerve section; n.a. � notavailable. ↑ � higher than all CIAP and autopsy controls;↓ � lower than all CIAP and autopsy controls.

were taken together, there was no significant differencebetween CIDP and CIAP.

In the qualitative analysis of teased fibre preparations bytwo independent neuropathologists, neither neuropathologistwas able to assign the diagnosis CIDP or CIAP based onqualification of the teased fibres (P � 1.00): oneneuropathologist assigned the term ‘mainly demyelinating’to seven preparations of 14 patients with CIDP, and threepreparations of nine patients with CIAP. The secondneuropathologist assigned the term ‘mainly demyelinating’to eight preparations of the 14 patients with CIDP, and fourpreparations of the nine patients with CIAP. Of the ninepreparations of patients with CIAP, two were labelled normalby the first neuropathologist and three by the secondneuropathologist. The agreement between the twoneuropathologists was moderate for CIDP (κ � 0.57) andlow for CIAP (κ � 0.18).

Onion bulbsThere was no significant difference in the number of onionbulbs between patients with CIDP and CIAP, but in bothpatient groups the numbers of onion bulbs were significantlyhigher than in autopsy controls (Fig. 2B). Four patients withCIDP had a higher number of onion bulbs than any of theCIAP and autopsy controls (Fig. 2B and Table 4).

g ratioIn patients with CIDP, the g ratio was significantly lowerthan in patients with CIAP and autopsy controls (Fig. 2C).Only one patient with CIDP had a lower g ratio than any ofthe CIAP and autopsy controls (Fig. 2C, Table 4).

Axonal de- and regenerationMyelinated nerve fibre densityMyelinated nerve fibre densities did not differ betweenpatients with CIDP and CIAP, but were significantly lowerthan autopsy controls (Fig. 4A). Three patients with CIDPhad higher and two lower myelinated nerve fibre densitiesthan any of the CIAP or autopsy controls (Fig. 4A).

ClustersThe number of clusters was not increased in CIDP comparedwith autopsy controls, whereas in CIAP the number ofclusters was significantly higher than in CIDP (Fig. 4B). Onepatient with CIDP had a lower number of clusters than anyof the CIAP and autopsy controls (Fig. 4B).

Internode lengthThe internode length was significantly larger in patients withCIDP than in patients with CIAP (Fig. 4C). Ten patientswith CIDP had larger internode lengths than CIAP controls(Fig. 4C).

InflammationEndoneurial areaThe endoneurial area in CIDP was not significantly differentfrom that in CIAP and autopsy controls (Fig. 5A). Fivepatients with CIDP showed a higher endoneurial area thanany of the controls (Fig. 5A and Table 4).

T cellsNumbers of T cells in the total sural nerve area weresignificantly higher in CIDP than CIAP or autopsy controls.Six out of 23 patients with CIDP showed a higher numberof T cells than any of the CIAP and autopsy controls [Fig.5B, adapted from our previous study (Bosboom et al., 1999)].In the endoneurium, the number of T cells was significantlyhigher than in CIAP or autopsy controls. Four patients withCIDP had an increased number of T cells compared with theCIAP and autopsy controls [Fig. 5C, adapted from ourprevious study (Bosboom et al., 1999)].

Histograms of myelinated nerve fibre diametersand axon diametersThe histograms of the myelinated nerve fibre diameters wereunimodal in both CIDP and CIAP (Fig. 6). There were

2434 W. M. J. Bosboom et al.

Fig. 3 Teased fibres of sural nerve biopsy specimens, classified according to Dyck (Dyck et al., 1993) inpatients with CIDP and CIAP. P � P value of the Mann–Whitney U-test used to compare CIDP withCIAP patients.

Fig. 4 The myelinated nerve fibre density (A), number of clusters (B) and internode length (C) in sural nerve biopsy specimens ofpatients with CIDP and CIAP, and normal controls. Normal � autopsy controls; *,** � statistically different from patients with CIDP(*P � 0.05, **P � 0.01); long horizontal lines represent the upper and lower limits for CIAP and autopsy controls; the short horizontalline represents the median.

significantly more large myelinated nerve fibres in patientswith CIDP than in those with CIAP. The histograms of theaxon diameter of the myelinated nerve fibres were unimodalin patients with CIDP and CIAP, and in autopsy controls. InCIDP, significantly more fibres with a small diameter werefound than CIAP or autopsy controls. In both CIDP andCIAP, a significant decrease in the largest axon diameterswas found compared with autopsy controls.

Multivariate analysesIt was possible that age could influence the results ofpathological features in the sural nerve biopsy specimens

(Berthold et al., 1983; Jacobs and Love, 1985). However,when comparisons were adjusted for age, all significantresults remained significant, and all non-significant resultsremained non-significant. The time from onset of diseaseuntil biopsy was significantly shorter in CIDP than CIAP(P � 0.02), and therefore we adjusted the comparisons ofthe quantitative analyses between CIDP and CIAP for timefrom onset of disease to biopsy: the differences in the numberof clusters and internode length between CIDP and CIAPwere no longer significant (P � 0.07 and P � 0.11). It wasalso possible that the myelinated nerve fibre density mightbe influenced by endoneurial oedema, but the difference in

Sural nerve demyelination in CIDP 2435

Fig. 5 The endoneurial area (A), number of T cells in the total sural nerve area (B) and in the endoneurium (C) of sural nerve biopsyspecimens of patients with CIDP and CIAP, and normal controls. Normal � autopsy controls; *,** � statistically different from patientswith CIDP, *P �0.05, **P� 0.01; long horizontal lines represent the upper and lower limits for CIAP and autopsy controls; the shorthorizontal line represents the median.

myelinated nerve fibre density remained significant afteradjustment for endoneurial area.

In both the univariate and the multivariate analyses offeatures in the transverse sections of all sural nerve biopsyspecimens, differences in g ratio between CIDP and CIAPgave the most differentiation between CIDP and CIAP. Teasedfibre preparations of the sural nerve biopsy specimen wereavailable in a subgroup of patients with CIDP and CIAP. Inthe multivariate regression analyses of all features, includingthe internode length, which could only be performed in thissubgroup of patients, the internode length differentiated themost between CIDP and CIAP.

Demyelinating and inflammatory features inindividual patients with CIDPTable 4 shows the individual CIDP patients with increasedor decreased values of demyelinating or inflammatoryfeatures. In 12 out of 21 patients with CIDP, one or morefeatures had increased values (i.e. higher than the highestvalue in patients with CIAP and in autopsy controls).

Correlation of pathological features in patientswith CIDPIn CIDP, a small axon diameter was correlated with a smallg ratio (R � 0.70, P � 0.01). A higher endoneurial area wascorrelated with a higher T-cell number in the total sural nervearea (R � 0.50, P � 0.03) and in the endoneurium (R �0.56, P � 0.01). A higher number of clusters was correlatedwith a lower T-cell density (R � 0.49, P � 0.04).

Correlation of clinical and pathological featuresin patients with CIDPIn the CIDP patients, there was no correlation between theseverity of the disease (maximal disability score) and any of

the pathological features. Older age was correlated withlower myelinated nerve fibre densities (R � 0.56, P � 0.01)and smaller internode lengths (R � –0.60, P � 0.02). Patientswith a higher CSF protein had a more rapid disease course(R � 0.54, P � 0.01) and a lower myelinated nerve fibredensity (R � 0.54, P � 0.01). Patients with the most severeoutcome had smaller axon diameters (R � 0.51, P � 0.03).

DiscussionTo determine whether light microscopical analysis of suralnerves provides diagnostic tools for the diagnosis of CIDP,we analysed sural nerve biopsy specimens of 21 patientswith CIDP. Abnormalities were compared with sural nervebiopsy specimens of 13 patients with CIAP and six autopsycontrols. Patients with CIAP were chosen as controls, asthe polyneuropathy in this condition is based upon axonaldegeneration and has no inflammatory signs. Although somesignificant differences were found in CIDP compared withcontrols, there was a considerable overlap, limiting thediagnostic value in individual patients. In CIDP, no correlationwas found between severity of the disease and lightmicroscopical abnormalities.

In previous studies of CIDP, segmental de- andremyelination ranged from 19 to 77% of the teased fibres inone study, and was found in 88% of the patients in anotherstudy (Matsumuro et al., 1994; Bouchard et al., 1999).However, neither study compared the extent of de- andremyelination in CIDP with other neuropathies or controlswithout neuropathy. Krendel and colleagues compared suralnerve biopsy specimens of CIDP patients with those ina heterogeneous control group. They found predominantdemyelination and onion bulb formation in CIDP and otherdemyelinating neuropathies, in contrast to axonalneuropathies (Krendel et al., 1989). However, segmental de-

2436 W. M. J. Bosboom et al.

Fig. 6 Histograms of the myelinated nerve fibre diameter and the axon diameter in sural nerve biopsy specimens of patients with CIDPand CIAP, and normal controls. ↑ � significantly higher than in patients with CIDP; ↓ � significantly lower than in patients with CIDP.

and remyelination can also occur in axonal neuropathies(Jennekens et al., 1969; Llewelyn et al., 1991). Moreover, arecent study demonstrated that nerve biopsies did not helpto differentiate between CIDP and diabetic polyneuropathy,as segmental de- and remyelination, onion bulbs andinflammatory infiltrates were present in both (Uncini et al.,1999). The onion bulbs we observed in axonal neuropathymight be pseudo onion bulbs derived from axonaldegeneration and regeneration (Midroni and Bilbao, 1995).In our light microscopic study, we found no differences in

the aspect of onion bulb-like structures between CIDP andCIAP. Our systematic analysis did not show significantdifferences in demyelinating features between patients withCIDP and CIAP. The finding of characteristic demyelinatingchanges, such as large onion bulbs and naked axons,described in previous studies on sural nerve biopsyspecimens of patients with CIDP (Prineas, 1971; Prineasand McLeod, 1976; Rizzuto et al., 1982), may have helpedto confirm the diagnosis. However, we did not find thesefeatures frequently in our patients with CIDP. Demyelinating

Sural nerve demyelination in CIDP 2437

features were only prominent in a minority of theCIDP patients.

As the g ratio is determined by the ratio of axon to fibrediameters, a higher g ratio can be expected in a demyelinatingneuropathy, as remyelination leads to thinner myelin sheathswhereas the axon diameter remains the same. However, weobserved lower g ratios in CIDP than in CIAP or autopsycontrols. This finding can be explained by four mechanisms.(i) Shrinkage of axons in CIDP. Reduction of axon diametershas been observed to occur at an early stage of thedemyelinating process before much destruction of the myelinsheath has occurred (Prineas and McLeod, 1976), and topersist for months during remyelination (Rizzuto et al., 1982;Gabreels-Festen et al., 1992). Our axon diameter histogramsindicate that large fibres were lost or had shrunk in bothCIDP and CIAP when compared with autopsy controls. Inour total myelinated nerve fibre diameter histograms, therewas only a decrease of large fibres in CIAP. This couldindicate shrinkage of large axons in CIDP and loss of largefibres due to axonal degeneration in CIAP. (ii) Swelling ofmyelin in CIDP. Binding of antibodies or other humoralfactors to myelin may result in alteration of membranepermeability with subsequent intracellular Schwann celloedema (Sabatelli et al., 1996). Intramyelinic oedema hasbeen described in CIDP patients, with thinning or dis-appearance of myelin sheaths (Sabatelli et al., 1996). (iii) Thepresence of regenerated fibres with a thin myelin sheath inCIAP, resulting in higher g ratios. (iv) Loss of smallmyelinated nerve fibres in CIDP, as small myelinated nervefibres have higher g ratios than large myelinated nerve fibres(Friede and Beuche, 1985; Jacobs and Love, 1985). However,this hypothesis is not supported by our histograms. Althoughthe differences in g ratios between CIDP and CIAP areinteresting, only one patient with CIDP had a g ratio outsidethe range of the patients with CIAP. Therefore, g ratioanalysis does not discriminate between CIDP and CIAP inindividual patients.

Loss of myelinated nerve fibres is frequent in both axonalpolyneuropathy and CIDP. In CIDP, myelinated nerve fibreloss is correlated with the extent of de- and remyelination inteased fibre analysis as well as with the duration of thedisease, and the outcome (Bouchard et al., 1999; Nagamatsuet al., 1999). In addition, we demonstrated that myelinatednerve fibre loss in CIDP was correlated with higher CSFprotein levels and a more rapid disease course. Although inthe present study myelinated nerve fibre densities werereduced equally in CIDP and CIAP, in the latter conditionregeneration of myelinated nerve fibres seems moreprominent, as was demonstrated by increased numbers ofclusters and shortened internode lengths. This finding can beexplained partly by a longer time from onset of disease tothe moment of biopsy in CIAP than in CIDP.

Inflammatory features were not prominent in most patientswith CIDP. This study showed no consistent evidence ofendoneurial oedema in CIDP. In our previous study, T cellswere found in sural nerve biopsy specimens of all patients

with CIDP and CIAP, and also in autopsy controls. In CIDP,increased numbers of sural nerve T cells were only found ina minority of patients (Bosboom et al., 1999).

At least three reasons can be given as to why featuresof demyelination and inflammation in sural nerve biopsyspecimens did not differ between CIDP and CIAP. First, inCIDP, pathological changes are more extensive in proximalportions of the peripheral nerves. Secondly, the sural nerveis a sensory nerve, and motor signs and symptoms in CIDPare often more prominent than sensory signs (Hall et al.,1992; Dyck et al., 1993a). Thirdly, at the time of biopsy,disease-specific features may already have disappeared.

A previous retrospective clinical study showed noadditional value of the sural nerve biopsy for the diagnosisof CIDP as the diagnosis did not change after examinationof the sural nerve biopsy (Molenaar et al., 1998). However,a recent prospective study on the usefulness of sural nervebiopsy in various axonal and demyelinating neuropathiesshowed an alteration of the diagnosis in 14% of the casesafter sural nerve biopsy (Gabriel et al., 2000). The presentstudy, using systematic light microscopic analysis of suralnerves, demonstrates that features of demyelination andinflammation in the majority of patients with CIDP were notdifferent from those in controls. Therefore, a sural nervebiopsy in CIDP is of less diagnostic value than suggested byprevious studies.

AcknowledgementsWe wish to thank Professor F. G. I. Jennekens, MD, and G.H. Jansen, MD, for their expert analysis of teased fibres,Professor F. G. I. Jennekens for advice on the preparation ofthe manuscript, C. W. A. M. Engels for laboratory assistance,and C. J. M. Klijn for help with the statistical analyses.This work was supported by grants from the NetherlandsOrganization for Scientific Research, the Prinses BeatrixFonds and the Kroger Foundation, and the research of Dr Vanden Berg was supported by a fellowship from the RoyalNetherlands Academy of Arts and Sciences.

ReferencesBarohn RJ, Kissel JT, Warmolts JR, Mendell JR. Chronicinflammatory demyelinating polyradiculoneuropathy. Clinicalcharacteristics, course, and recommendations for diagnostic criteria.Arch Neurol 1989; 46: 878–84.

Berthold CH, Nordborg C, Hildebrand C, Conradi S, Sourander P,Lugnegard H. Sural nerve biopsies from workers with a history ofchronic exposure to organic solvents and from normal control cases.Morphometric and ultrastructural studies. Acta Neuropathol (Berl)1983; 62: 73–86.

Bosboom WM, Van den Berg LH, De Boer L, Van Son MJ, VeldmanH, Franssen H, et al. The diagnostic value of sural nerve T cells inchronic inflammatory demyelinating polyneuropathy. Neurology1999; 53: 837–45.

2438 W. M. J. Bosboom et al.

Bouchard C, Lacroix C, Plante V, Adams D, Chedru F, GuglielmiJM, et al. Clinicopathologic findings and prognosis of chronicinflammatory demyelinating polyneuropathy. Neurology 1999; 52:498–503.

Cornblath DR, Asbury AK, Albers JW, Feasby TE, Hahn AF,McLeod JG, et al. Research criteria for diagnosis of chronicinflammatory demyelinating polyneuropathy (CIDP). Report froman Ad Hoc Subcommittee of the American Academy of NeurologyAIDS Task Force. [Review]. Neurology 1991; 41: 617–18.

Dyck PJ, Prineas J, Pollard J. Chronic inflammatory demyelinatingpolyradiculoneuropathy. In: Dyck PJ, Thomas PK, Griffin JW, LowPA, Poduslo JF, editors. Peripheral neuropathy. 3rd edn. Philadelphia:W. B. Saunders; 1993a. p. 1498–517.

Dyck PJ, Giannini C, Lais A. Pathologic alterations of nerves. In:Dyck PJ, Thomas PK, Griffin JW, Low PA, Poduslo JF, editors.Periperal neuropathy. 3rd edn. Philadelphia: W. B. Saunders; 1993b.p. 514–95.

Friede RL, Beuche W. Combined scatter diagrams of sheaththickness and fibre calibre in human sural nerves: changes with ageand neuropathy. J Neurol Neurosurg Psychiatry 1985; 48: 749–56.

Gabreels-Festen AA, Joosten EM, Gabreels FJ, Jennekens FG,Janssen-van Kempen TW. Early morphological features indominantly inherited demyelinating motor and sensory neuropathy(HMSN type I). J Neurol Sci 1992; 107: 145–54.

Gabriel CM, Howard R, Kinsella N, Lucas S, McColl I, SaldanhaG, et al. Prospective study of the usefulness of sural nerve biopsy.J Neurol Neurosurg Psychiatry 2000; 69: 442–6.

Hall SM, Hughes RA, Atkinson PF, McColl I, Gale A. Motor nervebiopsy in severe Guillain–Barre syndrome. Ann Neurol 1992; 31:441–4.

Jacobs JM, Love S. Qualitative and quantitative morphology ofhuman sural nerve at different ages. Brain 1985; 108: 897–924.

Jennekens FG, Van der Most Van Spijk D, Dorhout Mees EJ. Nervefibre degeneration in uraemic polyneuropathy. Proceedings of theEuropean Dialysis and Transplant Association; 1969. p. 191–7.

Krendel DA, Parks HP, Anthony DC, St. Clair MB, GrahamDG. Sural nerve biopsy in chronic inflammatory demyelinatingpolyradiculoneuropathy. Muscle Nerve 1989; 12: 257–64.

Llewelyn JG, Gilbey SG, Thomas PK, King RH, Muddle JR,Watkins PJ. Sural nerve morphometry in diabetic autonomic andpainful sensory neuropathy. A clinicopathological study. Brain 1991;114: 867–92.

Low PA, McLeod JG, Prineas JW. Hypertrophic Charcot–Marie–Tooth disease. Light and electron microscope studies of the suralnerve. J Neurol Sci 1978; 35: 93–115.

Matsumuro K, Izumo S, Umehara F, Osame M. Chronicinflammatory demyelinating polyneuropathy: histological andimmunopathological studies on biopsied sural nerves. J Neurol Sci1994; 127: 170–8.

McCombe PA, Pollard JD, McLeod JG. Chronic inflammatorydemyelinating polyradiculoneuropathy. A clinical andelectrophysiological study of 92 cases. Brain 1987; 110: 1617–30.

Midroni G, Bilbao JM. Biopsy diagnosis of peripheral neuropathy.Boston: Butterworth–Heinemann; 1995. p. 45–74.

Molenaar DS, Vermeulen M, de Haan R. Diagnostic value of suralnerve biopsy in chronic inflammatory demyelinating polyneuropathy.J Neurol Neurosurg Psychiatry 1998; 64: 84–9.

Nagamatsu M, Terao S, Misu K, Li M, Hattori N, Ichimura M,et al. Axonal and perikaryal involvement in chronic inflammatorydemyelinating polyneuropathy. J Neurol Neurosurg Psychiatry 1999;66: 727–33.

Notermans NC, Wokke JH, Franssen H, van der Graaf Y, VermeulenM, Van den Berg LH, et al. Chronic idiopathic polyneuropathypresenting in middle or old age: a clinical and electrophysiologicalstudy of 75 patients [see comments]. J Neurol Neurosurg Psychiatry1993; 56: 1066–71.

Notermans NC, Wokke JH, van der Graaf Y, Franssen H, van DijkGW, Jennekens FG. Chronic idiopathic axonal polyneuropathy: afive year follow up. J Neurol Neurosurg Psychiatry 1994; 57: 1525–7.

Notermans NC, Wokke JH, van den Berg LH, van der GraafY, Franssen H, Teunissen LL, et al. Chronic idiopathic axonalpolyneuropathy. Comparison of patients with and withoutmonoclonal gammopathy. Brain 1996; 119: 421–7.

Prineas JW. Demyelination and remyelination in recurrent idiopathicpolyneuropathy. An electron microscope study. Acta Neuropathol(Berl) 1971; 34.

Prineas JW, McLeod JG. Chronic relapsing polyneuritis. J NeurolSci 1976; 27: 427–58.

Rizzuto N, Moretto G, Monaco S, Martinelli P, Pazzaglia P. Chronicrelapsing polyneuritis. Acta Neuropathol (Berl) 1982; 56: 179–86.

Sabatelli M, Mignogna T, Lippi G, Porcu C, Tonali P. Intramyelinicedema in chronic inflammatory demyelinating polyneuropathy. ClinNeuropathol 1996; 15: 17–21.

Teunissen LL, Notermans NC, Franssen H, van der Graaf Y, OeyPL, Linssen WH, et al. Differences between hereditary motor andsensory neuropathy type 2 and chronic idiopathic axonal neuropathy.A clinical and electrophysiological study. Brain 1997; 120: 955–62.

Uncini A, De Angelis MV, Di Muzio A, Callegarini C, Ciucci G,Antonini G, et al. Chronic inflammatory demyelinatingpolyneuropathy in diabetics: motor conductions are important inthe differential diagnosis with diabetic polyneuropathy. ClinNeurophysiol 1999; 110: 705–11.

Van Swieten JC, Koudstaal PJ, Visser MC, Schouten HJ, van GijnJ. Interobserver agreement for the assessment of handicap in strokepatients. Stroke 1988; 19: 604–7.

Received November 24, 2000. Revised April 12, 2001.Accepted July 10, 2001