Multiple sclerosis (MS) is characterised by inflammation, demyelination and gliosis, involving the central nervous system (CNS) and sparing the peripheral nervous system (PNS). Chronic inflammatory demyelinating polyneuropathy (CIDP) is regarded as being restricted to the PNS as the immune response seems to be directed against an antigen confined to PNS proteins.1 However, patients presenting both central and peripheral demyelination have occasionally been reported, and might therefore represent a distinct subpopulation. Common immunopathogenic mechanisms may underlie both the central and peripheral myelin involvement.2 3 We report five patients presenting a recurrent demyelinating disease fulfilling the criteria both for MS and CIDP. The homogeneity of the clinical and laboratory data of most of these patients suggests an immunopathogenesis distinct from that of classical MS and CIDP.

METHODS

This is a study of the clinical and laboratory data of five patients presenting with demyelinating disease affecting both the CNS and PNS. For each patient the study noted age at onset of the neurological signs and the dates that CNS and PNS involvement began (table 1).

The clinical and paraclinical workup for all patients included the following: brain and spinal cord MRI, CSF examination with isoelectrofocusing to evaluate oligoclonal bands, visual evoked potentials (VEPs) except for the first patient, electromyography (EMG) and biological measurements to verify the absence of metabolic causes (for diabetes mellitus, dysthyroidism, vitamin deficiency, plasma long chain fatty acids, phytanic acid, arylsulphatase and α and β galactocerebrosidase measurements), inflammatory or dysimmune causes (haemogram, C reactive protein, sedimentation velocity, antinuclear antibodies, cryoglobulinaemia, rheumatoid factor, antiphospholipid syndrome, serum immunoelectrophoresis, angiotensin conversion enzyme) or infectious causes (Lyme, syphilis, brucellosis, HIV, and B and C hepatitis serologies were negative). Paraclinical findings were completed with pulmonary radiography, accessory salivary gland biopsies and skin examination Antiganglioside antibodies were measured and, in particular, anti-myelin associated glycoprotein (anti-MAG) activity. Duplications or deletions of PMP 22, P0 and connexin 32 were examined in all patients. A peripheral nerve biopsy was performed in two patients (patient Nos 2 and 4) and a brain biopsy was also performed in patient No 4. For MS, the study applied the revised McDonald criteria and for CIDP the Ad Hoc Subcommittee criteria, Nicolas et al’s criteria and the criteria of the European Federation of Neurological Society (EFNS).4^–7

RESULTS

Clinical data for the five patients (four men and one woman) are summarised in table 1, laboratory findings in table 2 and EMG findings in table 3.

In 1989, patient No 1, a 21-year-old man, presented with sensory disturbances, motor dysfunction in the lower limbs and bladder impairment, and was initially diagnosed with MS. CSF was normal, without oligoclonal bands. The same year, distal paraesthesias of the lower limbs led to an EMG being performed, the results of which were normal. In 2001, depressed deep tendon reflexes and a proximal deficit in the lower limbs were noted. In 2003, a second EMG revealed a severe demyelinating neuropathy without conduction blocks or dispersion. No evidence of metabolic disorders, inflammatory, dysimmune or infectious causes or vaccinations was found.

In 2002, patient No 2, a 38-year-old woman, was diagnosed with MS. She presented with recurrent episodes of ataxia, dysgueusia and optic neuritis. Her brain MRI suggested MS, with numerous periventricular and subcortical T2 hyperintense lesions fulfilling Barkhof’s criteria for spatial dissemination (fig 1A).8 She was treated with interferon β1a (IFNβ1a) in 2002, then IFNβ1b in 2004. In 2005, she presented with aphasia, diplopia and severe ataxia, and a major proximal deficit in the lower limbs. On brain MRI, numerous gadolinium enhanced tumefactive lesions were observed in white and grey matter (fig 1B, C). Spinal cord MRI showed transverse myelitis with gadolinium enhanced roots (fig 1D). Although deep tendon reflexes were preserved and brisk, the EMG in 2005 revealed a severe sensorimotor demyelinating neuropathy without conduction blocks or dispersion. CSF analysis showed high hyperproteinorachia (3.39 g/l) without cells and with no oligoclonal bands. No evidence of metabolic disorders, inflammatory, dysimmune or infectious causes or vaccinations was found. Peripheral nerve biopsy showed mononuclear cell infiltrates, endoneural oedema, demyelination and remyelination with thinly myelinated axons, without vasculitis or abnormal cells (fig 2). Clinical signs and MRI and EMG parameters improved after immunosuppressive therapy with monthly intravenous cyclophosphamide, but she remained severely disabled.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Brain and spinal cord MRI of patient No 2. (A) T2 fluid attenuated inversion recovery (FLAIR) weighted brain MRI in the axial plane in 2002, showing periventricular and subcortical T2 hyperintense lesions. (B, C) Brain MRI in T2 FLAIR weighted sequence (B) and T1 weighted sequence (C) in the axial plane in 2005, showing pre-rolandic tumefactive T2 hyperintensity lesions in grey and white matter (B), which were gadolinium enhanced (C). (D) T1 weighted spinal cord MRI in the sagittal plane in 2005, showing gadolinium enhanced caudal roots (white arrows).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 (A) Section of the sural nerve biopsy showing loss of myelinated fibres, thin myelin sheaths and endoneural oedema (arrow). (B) Electron microscopic longitudinal section of the sural nerve biopsy showing demyelination of one node of Ranvier (black arrow). Scale bars: (A) 1500 mm; (B) 2 μm.

In 1997, patient No 3, a 29-year-old man, presented with facial and perioral paraesthesia and diarrhoea. In 2002, he complained of lumbar pain and a predominantly proximal sensorimotor deficit in the lower limbs, which was associated with perioral paraesthesia and agueusia. Neurological examination revealed brisk reflexes, no Babinski sign and distal hypoesthesia in the lower limbs. Brain MRI fulfilled Barkhof’s criteria, with numerous periventricular T2 hyperintensity lesions (fig 3A, B) and gadolinium enhanced lesions on T1 weighted sequences (fig 3C). CSF analysis revealed high hyperproteinorachia (4 g/l) without cells or oligoclonal bands. Nerve conduction studies showed signs of demyelinating neuropathy, suggesting an atypical Guillain–Barré syndrome because of the lack of conduction block or dispersion and because of the CNS involvement. No evidence of metabolic disorders, inflammatory, dysimmune or infectious causes or vaccinations was found. The patient received 400 mg/kg/day of intravenous IgG for 5 days, which led to an improvement in sensory and motor deficits, and normalised EMG parameters. From 2003, he complained of numbness in the lower limbs, dizziness, dysarthria and bladder dysfunction. In 2005, he presented with visual disturbances, mild motor deficits in the lower limbs and left homonymous lateral hemianopsia with headaches. The history fulfilled McDonald’s criteria for MS, and brain MRI was also highly suggestive of MS.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Brain MRI of patient No 3. (A, B) T2 fluid attenuated inversion recovery weighted brain MRI in the axial plane, showing periventricular T2 hyperintense lesions. (C) T1 weighted brain MRI in the axial plane showing a gadolinium enhanced lesion (centre right).

From 1995 onwards, patient No 4, a 65-year-old man, presented with recurrent neurological episodes with bilateral optic neuritis, cerebellar disturbances and sensorimotor deficits in the lower limbs. Brain MRI showed multiple gadolinium enhanced lesions involving grey and white matter. The initial CSF analysis showed pleiocytosis with 7 lymphocytes/mm³ without hyperproteinorachia and with no oligoclonal bands. Brain biopsy revealed perivascular cells with macrophages in lesions. In 1997, areflexia and hypopallesthesia in all four limbs with predominantly proximal motor deficit led to an EMG being performed, which revealed a demyelinating neuropathy without conduction block or abnormal temporal compound muscle action potential dispersion. At that time, a second CSF analysis revealed a proteinorachia of 1 g/l without cells and without oligoclonal bands. No evidence of metabolic disorders, inflammatory, dysimmune or infectious causes or vaccinations was found. No antiganglioside antibodies were detectable and, in particular, no anti-MAG activity was evidenced. PMP 22 point mutation, duplication or deletion, and P0 and connexin 32 mutations were looked for and were negative. A neuromuscular biopsy showed demyelination and remyelination and some inflammatory infiltrates without vasculitis or abnormal cells. The disease course was marked by a recurrence of neurological disorders with aphasia, cognitive dysfunction and seizures, as well as with a recurrent proximal motor deficit in the lower limbs. In 2002, myelitis was demonstrated by spinal cord MRI. The patient fulfilled Barkhof’s criteria on MRI and McDonald’s criteria for MS diagnosis. Clinical signs and EMG parameters partially improved after corticosteroid and cyclophosphamide therapy but, thereafter, the patient’s condition progressively worsened. He became bedridden and developed severe cognitive dysfunction with a frontal syndrome.

In 1982, patient No 5, a 26-year-old man, was diagnosed with clinically definite relapsing–remitting MS and suffered relapses in 1984 and 1992. In 1999, he was treated with IFNβ1a. Deep tendon reflexes were brisk. CSF analysis was normal. Brain MRI fulfilled Barkhof’s criteria for spatial dissemination. Following the onset of thenarian amyotrophy in 2004 associated with areflexia, an EMG was performed, which revealed a demyelinating neuropathy without conduction block or dispersion. No evidence of metabolic disorders, inflammatory, dysimmune or infectious causes or vaccinations was found. The patient remained stable on IFNβ1a treatment.

None of the patients had a family history of neurological disorders. No evidence of metabolic disorders, inflammatory or infectious causes or vaccinations was found (plasma long chain fatty acids, phytanic acid, arylsulphatase and α and β galactocerebrosidase). No antiganglioside antibodies were detectable and, in particular, no anti-MAG activity was evidenced. PMP 22 point mutation, duplication or deletion, and P0 and connexin 32 mutations were negative for all patients.

DISCUSSION

We studied five patients with a recurrent history of neurological signs involving demyelinating processes in both the CNS and PNS. Our clinical and paraclinical data were compared with those found in the literature (table 4).

Although the five patients reported here fulfilled Barkhof’s criteria on MRI and the McDonald criteria for MS definition, we consider that the symptoms and MRI features are atypical of MS.4 22 Indeed, although MS lesions may involve grey matter and be potentially pseudotumoral, the extensive transverse myelitis in patient No 2 is atypical.23 Four patients presented with a high level of hyperproteinorachia, and none of the patients had oligoclonal bands in CSF, whereas oligoclonal bands are reported to be present in more than 90% of MS cases.24 The prevalence of PNS involvement in MS remains unknown.25 26 There is only one retrospective study analysing its frequency. In a cohort of 150 patients with MS, Zee et al reported clinical and neurophysiological signs of PNS involvement in 13 patients, including a demyelinating polyneuropathy in four patients.26 Demyelinating lesions may rarely occur in spinal roots and in peripheral nerves in MS. Two patients have previously been reported to have MS associated with symptomatic CIDP, with massive spinal or cranial nerve hypertrophy revealed by MRI.27 It was intriguing to observe that two of our patients (patient Nos 2 and 5) developed CIDP, as documented by EMG findings, while receiving IFNβ, in one case after 3 years of treatment and in the other case after 6 years. Recent reports have described three children with MS who developed CIDP during the course of IFNβ therapy (4 months, 1 year and 4 years, respectively, after the introduction of IFNβ therapy), suggesting that IFNβ does not protect against CIDP.17 Unlike our cases, these children presented with conduction blocks and temporal dispersion on their EMG findings (table 4). More recently, a case was reported of a woman with childhood onset relapsing–remitting MS who developed CIDP without conduction blocks or dispersion, 2 months after the introduction of treatment with IFNβ.18 As Merrigglioni and Rowin have also pointed out, it is widely recognised that patients with one autoimmune disease may be more susceptible to developing another, even without therapy with an immunomodulatory agent.27 Durelli et al found an increase in autoimmune diseases and autoantibodies in patients with MS treated with IFNβ.28 Moreover, although IFNa, like IFNβ, has been incriminated in CIDP development,29 it could have a potential effect in the treatment of CIDP.30

Some cases of connexin 32 and P0 mutation have been described in Charcot–Marie–Tooth disease associated with brainstem involvement and demyelinating lesions or brain magnetic resonance abnormalities.31 32 However, none of our patients presented PMP22, P0 or connexin 32 mutations.

Alternatively, it could be postulated that our patients had an acquired demyelinating polyneuropathy having similarities with CIDP, associated with CNS involvement. All but patient No 3 fulfilled the EFNS (2005) criteria and Nicolas et al’s neurophysiological criteria for CIDP diagnosis.6 7 Only one (patient No 4) of the five patients fulfilled the more stringent Ad Hoc Subcommittee neurophysiological criteria for CIDP.5 The distal motor latency retardation associated with motor conduction velocity slowing and the hyperproteinorachia in three other patients (patient Nos 1, 2 and 5) fulfilled Nicolas et al’s criteria for CIDP.6 Patient No 3 did not fulfil Nicolas et al’s criteria or the Ad Hoc Subcommittee’s criteria for CIDP, but the increased distal motor latency in the median and peroneal nerves associated with a very high level of hyperproteinorachia was consistent with CIDP according to the EFNS/PNS guidelines for chronic demyelinating polyradiculopathy. None of our patients had conduction block or abnormal temporal compound muscle action potential dispersion whereas most patients with CNS involvement described in the literature usually presented with conduction blocks (table 4). Three of our patients (patient Nos 2, 3 and 4) had a high level of proteinorachia (>1 g/l). CSF analysis in patient Nos 1 and 5 did not reveal hyperproteinorachia but it was performed before the occurrence of PNS involvement. Had the analysis been performed after the PNS signs were documented it might well have shown an increased protein level. CIDP is an acquired neuropathy in which CNS demyelinating lesions are not common but have been described.1 Based on MRI evidence, such lesions could be found in 3–23% of cases.27 33 34 All of our patients presented with CNS lesions. Among the 100 CIDP patients described by Bouchard et al, five had CNS clinical signs: three of them had MRI features of MS and two had oligoclonal bands in the CSF.35 Moreover, in an earlier study, we showed a high frequency of abnormal VEPs in patients with CIDP but found that they were poorly correlated with CNS demyelinating lesions.36 Among the 17 patients with CIDP in that study, eight had altered VEPs, but only four of these had MRI abnormalities.36 Arias et al reported an adult case of demyelinating neuropathy that was followed by a single CNS attack related to a pseudotumoral acute demyelinating lesion.13

There are some similarities between the proposed immunopathogenesis of CIDP and MS. In both cases the combination of a genetic predisposition and an environmental factor has been proposed as the trigger for the immune response.37 38 Reactive T cells and/or antibodies may mediate this immune response by crossing the blood–brain/blood–nerve barrier and activating an inflammatory process after being re-exposed to the autoantigen.30 31 Whether the same antigenic target occurs in both the PNS and CNS or whether there is cross reactivity between them or involving a secondary inflammatory reaction remains to be determined. It has been hypothesised that peripheral myelin P1 protein expressed in peripheral nerve is identical to myelin basic protein.39 The response to immunotherapy suggests a dysimmune process. The CNS involvement is highly suggestive of MS relapses. The PNS disorders in our patients occurred after the onset of CNS involvement and corresponded to a demyelinating neuropathy without conduction block or dispersion. Two MS patients whose MRI showed spinal and cranial hypertrophic neuropathy had demyelinating neuropathy without conduction block and their CSF showed no oligoclonal bands.20 In 2003, Rodriguez-Casero et al described a case of childhood onset MS associated with a demyelinating neuropathy.18 In all similar cases reported in the literature, no oligoclonal bands were found in CSF (table 4).

Lastly, our cases differed from acute disseminated encephalomyelitis, which is usually considered to be a monophasic demyelinating disease of the CNS that develops after an infection or a vaccination. Acute disseminated encephalomyelitis rarely involves the PNS and is usually associated with a lymphocytic pleiocytosis, which was not observed in our patients.40

In conclusion, we consider that these case reports correspond to a new entity with initiation of the inflammatory demyelinating disease in the CNS and its subsequent extension to the PNS. We hypothesise a dysimmune entity, distinct from classical MS and CIDP, with a demyelinating neuropathy without conduction block. Inflammatory and dysimmune processes in the CNS, differing from those occurring in typical MS, could lead to a breaching of the blood–brain/blood–nerve barrier and/or reveal putative hidden cryptic targets mimicking PNS myelin autoantigens.