Article Text

Download PDFPDF

Research paper
Uniform demyelination and more severe axonal loss distinguish POEMS syndrome from CIDP
  1. Michelle L Mauermann1,
  2. Eric J Sorenson1,
  3. Angela Dispenzieri2,
  4. Jay Mandrekar3,
  5. Guillermo A Suarez1,
  6. Peter J Dyck1,
  7. P James B Dyck1
  1. 1Department of Neurology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA
  2. 2Department of Hematology, Mayo Clinic College of Medicine, Rochester, Minnesota, USA
  3. 3Division of Biomedical Statistics and Informatics, Mayo Clinic College of Medicine, Rochester, Minnesota, USA
  1. Correspondence to Dr M L Mauermann, Department of Neurology, Mayo Clinic College of Medicine, 200 First Street SW, Rochester, MN 55905, USA; mauermann.michelle{at}


Objective POEMS syndrome (the acronym reflects the common features: Polyneuropathy, Organomegaly, Endocrinopathy, Monoclonal protein and Skin changes) is a paraneoplastic disorder with a ‘demyelinating’ peripheral neuropathy that is often mistaken for chronic inflammatory demyelinating polyradiculoneuropathy (CIDP). The nerve conduction study (NCS) and electromyography (EMG) attributes that might differentiate POEMS from CIDP and lead to earlier therapeutic intervention were explored.

Methods NCS/EMG of POEMS patients identified through retrospective review from 1960 to 2007 were compared with matched CIDP controls.

Results 138 POEMS patients and 69 matched CIDP controls were compared. POEMS patients demonstrated length dependent reduction in compound muscle action potentials, low conduction velocities, prolonged distal latencies and prolonged F wave latencies. Compared with CIDP controls, POEMS patients demonstrated: (1) greater reduction of motor amplitudes, (2) greater slowing of motor and sensory conduction velocities, (3) less prolonged motor distal latencies, (4) less frequent temporal dispersion and conduction block, (5) no sural sparing, (6) greater number of fibrillation potentials in a length dependent pattern and (7) higher terminal latency indices (TLI). TLI ≥0.38 in the median nerve demonstrated a sensitivity of 70% and specificity of 77% in discriminating POEMS from CIDP.

Conclusions NCS/EMG of POEMS syndrome suggests both axonal loss and demyelination. Compared with CIDP, there is greater axonal loss (reduction of motor amplitudes and increased fibrillation potentials), greater slowing of the intermediate nerve segments, less common temporal dispersion and conduction block, and absent sural sparing. These findings imply that the pathology of POEMS syndrome is diffusely distributed (uniform demyelination) along the nerve where the pathology of CIDP is probably predominantly proximal and distal. Median motor TLI may be useful in clinically distinguishing these disorders.

View Full Text

Statistics from


POEMS syndrome is a rare paraneoplastic disorder secondary to a plasma cell dyscrasia. The acronym POEMS reflects the common features: Polyneuropathy, Organomegaly, Endocrinopathy, Monoclonal protein and Skin changes.1 Other recognised traits include sclerotic bone lesions, extravascular volume overload, angiofollicular lymph node hyperplasia, papilloedema and thrombocytosis/polycythemia. Plasma and serum vascular endothelial growth factor (VEGF) levels are markedly elevated2–4 and correlate with disease activity.2 5 6

Peripheral neuropathy is usually the dominant feature of POEMS syndrome. The neuropathy often begins in the feet with positive neuropathic sensory symptoms followed by motor symptoms which progress in a length dependent fashion.7 Weakness becomes severe causing many patients to become wheelchair or bedbound.7 8 The neuropathy is often mistaken for chronic inflammatory demyelinating polyradiculoneuropathy (CIDP) as both present as subacute motor predominant polyradiculoneuropathies. Both share other features, such as albuminocytological dissociation and demyelinating features on electrophysiology.7 POEMS patients have other features suggesting more systemic illness, but these are commonly overlooked. Differentiation of POEMS syndrome from CIDP is useful as POEMS patients do not respond or respond poorly to conventional immunotherapy used for CIDP (plasmapheresis, intravenous immune globulin, oral steroids) and require radiation, systemic chemotherapy or high dose chemotherapy with stem cell transplant. In recent years, POEMS syndrome has become potentially curable through autologous stem cell transplant.

Some aspects of the electrophysiological features of POEMS syndrome have been investigated, although in very small series given the rarity that POEMS syndrome occurs. Given our large referral practice of patients with monoclonal protein disorders, and specifically POEMS syndrome, we wanted to further evaluate these findings in a large cohort using age and gender matched CIDP controls to determine if nerve conduction studies (NCS) and electromyography (EMG) can differentiate POEMS syndrome from CIDP and aid in understanding the pathophysiology of POEMS. We hypothesised that the NCS in POEMS syndrome would (1) demonstrate both axonal loss and demyelination, (2) confirm prior findings of more severe involvement of the lower limbs and (3) show less temporal dispersion and conduction block compared with CIDP.


Method of sampling

The study was approved by our institutional internal review board. The Mayo Clinic dysproteinaemia database was reviewed from 1 January 1960 to 31 July 2007 for the diagnosis of POEMS syndrome. Patients with CIDP were identified by retrospectively searching the Mayo Clinic database from the same period for CIDP, polyradiculoneuropathy, polyradiculitis and polyradiculopathy. The CIDP control patients were matched to POEMS patients for gender, date of birth within 1 year and date of evaluation within 1 year. Five CIDP controls were matched for each POEMS case.

Inclusion/exclusion criteria

Patients who met the criteria for POEMS syndrome9 were included. This included having both major criteria (polyneuropathy and monoclonal plasma cell disorder) and one minor criterion. A monoclonal plasma cell disorder is documented in any of three ways: (1) serum M protein; (2) positive biopsy of a plasmacytoma; (3) positive iliac crest biopsy. The minor criteria are a sclerotic bone lesion, Castleman's disease, organomegaly (splenomegaly, hepatomegaly or lymphadenopathy), oedema (pleural effusion or ascites), endocrinopathy (adrenal, thyroid, pituitary, gonadal, parathyroid, pancreatic), skin changes (hyperpigmentation, hypertrichosis, plethora, haemangiomata, white nails) or papilloedema. Due to the high prevalence of diabetes mellitus and thyroid abnormalities, these diagnoses alone were not sufficient to meet the minor criterion.

CIDP patients met clinical criteria from our recent epidemiology publication.10 The initial symptomatology and deficits had to worsen in a progressive or fluctuating course over a period of time ≥8 weeks. The course thereafter could be progressive, monophasic or relapsing. There had to be preferential or at least equal involvement of large nerve fibres compared with small sensory nerve fibres or autonomic involvement. The pattern of function involvement was that of a polyradiculoneuropathy with prominent limb involvement of the proximal or distal segments. Cases could be predominantly proximal or distal but had to have some degree of both to be included. There had to be motor weakness. The process had to be symmetric, with <25% difference between sides. There had to be hyporeflexia or areflexia. In the present study, CSF protein had to be >45 mg/dl and have a cell count of <10. If CSF was not done but the case met all of other criteria, it was included. The diagnosis was confirmed by the treating neurologist and one of the authors (MLM), independent of the NCS/EMG results. NCS/EMG data were not used for inclusion as the main study endpoint was NCS and the American Academy of Neurology criteria typically exclude about half of cases and leave only those with very demyelinating NCS.11 Both POEMS and CIDP patients had to have NCS/EMG performed at the time of initial evaluation at our institution.

Patients were excluded if they had neuropathies related to infectious or metabolic diseases, if there were diseases associated with nerve necrotising vasculitis (polyarteritis nodosa, Wegener's granulomatosis, rheumatoid arthritis, Churg–Strauss syndrome), microvasculitis (diabetic lumbosacral radiculoplexus neuropathy, lumbosacral radiculoplexus neuropathy), immune sensory neuropathies (Sjögren's, lupus) or granulomatous diseases (sarcoid, leprosy). CIDP patients were also excluded if there was a coexistent monoclonal protein, sclerotic bone lesion or other neoplastic disorder.

Nerve conduction studies

NCS were performed using surface stimulation and recording techniques (Teca TE42 and Nicolet Viking machines) standard for our laboratory.12 Skin temperatures were maintained above 32.0°C in the upper extremities and 30.0°C in the lower extremities.

Compound muscle action potentials (CMAPs) were evaluated from the peroneal (extensor digitorum brevis), tibial (abductor hallucis), median (abductor pollicis brevis) and ulnar (adductor minimi) nerves. Motor conduction velocities (MCV) were from wrist to elbow for the upper limb nerves and ankle to knee for lower limb nerves. F waves were recorded after distal stimulation and at least eight consecutive responses were analysed. The F estimates in the upper limbs were recorded using the formula: F estimate = (2 × the distance from the distal stimulation site to the sternal notch)/MCV + distal motor latency (DML). F corrected was calculated by subtracting the F estimate from the F wave latency. F waves were considered abnormal if ≥32 ms in the upper limb or ≥58 ms in the lower limb.

Sensory nerve action potentials (SNAPs) were recorded from the median, ulnar and sural nerves using antidromic studies. Sensory conduction velocities (SCV) were calculated from onset latency. The distal sensory latency (DSL) was recorded from the peak latency at the distal site of stimulation.

The presence or absence of conduction block and temporal dispersion were recorded. Conduction block was defined as either a ≥50% change in amplitude proximal:distal irrespective of distance or duration. In order to include only true conduction block, distal CMAP had to be ≥1 mV. Temporal dispersion was defined as duration of the distal potential/duration of the proximal potential <0.7. Studies performed prior to 1989 could not be assessed for temporal dispersion.

Terminal latency indices (TLI) were calculated for each motor study using the following formula: TLI=terminal distance (mm)/(distal sensory latency (ms) × MCV (m/s)).

Blink responses were performed by stimulating at the supraorbital notch and recording over the orbicularis oculi muscle on the inferior rim of the orbit directly below the centre of the pupil. The R1 and ipsilateral and contralateral R2 responses were recorded.

To assess the degree of nerve fibre loss and degree of demyelination in motor and sensory nerves, percentiles were used to compare summated values between POEMS and CIDP. Percentiles were calculated from the Rochester Diabetic Study-health subject cohort and corrected for anthropometric characteristics.13

Axonal involvement was also examined by needle EMG using a concentric needle. Summated abnormalities of fibrillation potentials, long duration motor unit potentials and recruitment pattern were abstracted from five muscle groups: tibialis anterior, vastus medialis/vastus lateralis/rectus femoris, first dorsal interosseous, thoracic paraspinals and lumbar paraspinals using standard grading.

Statistical analysis

Descriptive summaries are presented as frequencies and percentages for categorical variables and median and ranges for continuous variables. Comparisons between the POEMS and CIDP groups were performed using Fisher's exact test or Wilcoxon rank sum test, as appropriate. All of the tests were two sided, and p values <0.05 were considered statistically significant. As the TLI proved most discriminating, receiver operating curve (ROC) analysis on TLI was performed on a randomly selected half of the cases. From this half of the data, thresholds for distinguishing POEMS from CIDP were calculated and the sensitivity and specificity of these thresholds were tested using the second half of the cases.


From the POEMS database, 190 subjects were identified. Of these, 138 met all of the inclusion and exclusion criteria. Three subjects were excluded because of lack of neuropathy, 46 subjects because no NCS/EMG was performed at diagnosis at the Mayo Clinic and three subjects due to an alternative diagnosis of monoclonal gammopathy of undetermined significance (MGUS) or multiple myeloma. Age and gender CIDP matched controls were identified for 69 of the POEMS subjects. No matched CIDP controls could be identified for the other 87 POEMS cases. All 138 POEMS patients and 69 CIDP controls were included in the statistical analysis. Ninety POEMS patients (65%) and 43 CIDP controls (62%) were men (p=0.68). Median age of the POEMS patients was 51.5 years (range 20–73 years) and for CIDP controls, 47 years (range 19–74 years; p=0.04). Median weight of the POEMS patients was 76 kg (range 43–130) and 82.7 kg (range 53.6–138.6) for CIDP controls (p=0.01). Median body mass index and body surface area were greater in CIDP controls (p=0.003, p=0.03, respectively). There was no difference in median height (p=0.67) between the groups. Neuropathy impairment score was similar between the two groups, confirming a similar severity of neuropathy: POEMS 65 (4–168) points, CIDP 58 (6–155) points (p=0.22).

The clinical and laboratory characteristics of the POEMS patients are shown in table 1. All of these patients had a clonal plasma cell proliferative disorder, which was lambda in 97.1% of patients. Sclerotic bone lesions, endocrinopathies and skin changes were found in over half of the patients. Elevated CSF protein was present in 100% of patients tested. Of patients who had not been previously treated for POEMS syndrome, only one had a normal plasma VEGF.

Table 1

Clinical and laboratory characteristics of POEMS syndrome patients (n=138)

NCS parameters are shown in table 2. NCS were performed in POEMS patients on 462 motor nerves and 364 sensory nerves, and in CIDP controls on 226 motor nerves and 182 sensory nerves. POEMS patients demonstrated a reduction in both upper and lower limb CMAPs and SNAPs. CMAPs in the lower limbs were more severely affected than the upper limbs with abnormal values in 96.2% peroneal, 97.7% tibial, 65.2% ulnar and 67.2% median nerves tested. In general, the upper limb CMAP responses were present (absent responses: ulnar 5/135, median 9/31) whereas the lower limb responses were absent (absent responses: peroneal 108/133, tibial 104/133). The reduction in SNAPs was similar in the upper and lower limbs: sural 70.0%, median 75.7%, ulnar 72.7% (absent responses: sural 107/129, median 43/136, ulnar 42/99; absent sural over the age of 60 years was not considered abnormal). Compared with CIDP controls, POEMS patients demonstrated a greater reduction in CMAPs, especially in the lower limbs (peroneal p<0.0001, tibial p<0.001, median p=0.005). The sural SNAP also demonstrated a greater reduction in POEMS patients (p<0.0001).

Table 2

Nerve conduction study attributes in POEMS syndrome compared with CIDP controls

POEMS patients demonstrated reduced upper and lower limb motor and SCVs. MCVs were equally reduced in the upper and lower limbs. Abnormal values occurred in the ulnar (93.8%), median (98.1%), peroneal (96.2%) and tibial (93.3%) nerves. Compared with CIDP controls, POEMS patients demonstrated a greater reduction in MCV (ulnar p=0.009, median p=0.003, tibial p=0.03) and SCV (ulnar p=0.007, median p<0.001).

POEMS patients demonstrated prolongation of motor and sensory distal latencies. Abnormal values occurred in ulnar motor (68.5%), median motor (83%), tibial motor (50%), ulnar sensory (86%) and median sensory (79.6%). Compared with CIDP controls, POEMS patients demonstrated somewhat less distal latency prolongation (median DML p=0.049). Further evaluation of involvement of the distal nerve segment was evaluated using TLI which looks at the conduction of the distal segment compared with the intermediate segment. TLI showed more slowing of the distal motor nerve segment in CIDP controls compared with POEMS patients (ulnar p=0.001, median p=0.0005, tibial p=0.03).

POEMS patients demonstrated absolute prolongation of F wave latencies. Abnormal values occurred in the ulnar (95.5%), median (100%), peroneal (66.7%) and tibial (72.2%) nerves. There was greater prolongation compared with CIDP controls (ulnar p=0.01, median p=0.008, tibial p=0.046). However, differences in F corrected values were not significant. Blink responses demonstrated similar prolongation of the R1 latency in both POEMS and CIDP controls.

The degree of axonal loss was greater in POEMS patients compared with CIDP controls (table 3) (composite motor amplitude p=0.004, composite sensory amplitude p=0.002). The degree of demyelination was assessed by looking at both composite conduction velocities and composite distal latencies for motor and sensory nerves. All parameters demonstrated greater demyelination in POEMS patients compared with CIDP controls (MCV p<0.0001, SCV p<0.0001, DML p=0.02, DSL p<0.0001).

Table 3

Composite scores of axonal loss and demyelination

Conduction block was less frequent in POEMS patients compared with CIDP controls (6.7% vs 23.1%; p<0.0001) (table 4). Temporal dispersion was also less frequent in POEMS compared with CIDP controls (13.3% vs 17.4%; p=0.001).

Table 4

Presence of conduction block and temporal dispersion in POEMS syndrome compared with CIDP controls

The presence of sural sparing (present sural SNAP in the absence of median and ulnar SNAPs) was not seen in POEMS patients (0/30) compared with CIDP controls (5/10) (p=0.003).

EMG demonstrated primarily length dependent neurogenic abnormalities but there was evidence of proximal involvement consistent with a polyradiculoneuropathy (table 5). Compared with CIDP controls, POEMS patients demonstrated more fibrillation potentials in predominantly distal muscles (tibialis anterior p<0.0001, quadriceps p=0.03, first dorsal interosseous p=0.009) in a length dependent pattern. Neurogenic motor units were more commonly found in proximal muscles (thoracic paraspinals p=0.006) in CIDP controls.

Table 5

Needle electromyography findings in POEMS syndrome compared with CIDP controls

ROC analysis of the TLI for the ulnar, median, peroneal and tibial motor nerves was performed. This revealed an area under the curve of 0.74 for the median and 0.70 for the tibial motor nerves. ROC analysis of the ulnar and peroneal nerves failed to discriminate POEMS from CIDP with any accuracy. In the training half of the data, a TLI ≥0.38 in the median nerve and a TLI ≥0.43 in the tibial nerve were suggestive of POEMS. Applying this to the testing half of the cases revealed a sensitivity of 70% and specificity of 77% in discriminating POEMS from CIDP in the median motor nerve, and a sensitivity of 50% and specificity of 70% in the tibial motor nerve.


Our study describes the electrophysiological features of a large cohort of patients with POEMS syndrome. The series reported here is important because despite prior studies of small cohorts, POEMS syndrome remains underrecognised by neurologists and is often initially misdiagnosed as CIDP. Electrophysiological attributes that may heighten the examiner's suspicion of POEMS syndrome would be a helpful tool in improving diagnosis. Also, it is important to try to better understand the pathophysiological basis of the neurological involvement in POEMS syndrome.

Our study shows that the NCS attributes and needle EMG findings associated with POEMS syndrome suggest both a demyelinating and axonal polyradiculoneuropathy. Greater axonal loss in the lower limbs was seen in POEMS patients compared with CIDP, as evidenced by the greater reduction in CMAP and SNAP amplitudes. This finding of preferential involvement of the lower limbs was previously suggested by other studies of POEMS patients7 14 but comparison with CIDP controls was not done. In addition, needle EMG findings (fibrillation potentials and neurogenic motor units) in this study confirm that the loss of CMAP and SNAP amplitudes in POEMS syndrome is due to axonal loss rather than distal conduction block.

Demyelinating features include slowed conduction velocities, prolonged distal latencies, prolonged F wave latencies and R1 latencies of the blink response. These findings can also be seen in CIDP but there are differences that aid in distinguishing them. A previous study of eight patients with POEMS syndrome showed that there was less frequent prolongation of distal latencies but slowing in conduction velocities was similar to CIDP and Charcot Marie Tooth (CMT) patients, leading the authors to conclude that slowing was more prominent in intermediate nerve segments.15 We also found that intermediate conduction velocities were more slowed in POEMS than in CIDP, whereas distal latencies were somewhat more involved in CIDP. Combining the intermediate and distal nerve segments in the TLI is a useful way of distinguishing CIDP from POEMS syndrome. In addition, we found that F wave latencies compared with F estimates were similar and there was no significant difference from CIDP. Conduction block and temporal dispersion are often recognised as attributes that suggest an acquired rather than inherited form of peripheral demyelinating disease. A prior study reported less frequent conduction block in a cohort of eight POEMS patients.15 Our study demonstrated that only 16 of 238 POEMS NCS showed conduction block (6.7%) versus 46 of 199 CIDP controls (23.1%). We also found significantly less temporal dispersion in POEMS patients compared with CIDP. While the absence of conduction block and temporal dispersion are more suggestive of inherited demyelinating disorders, the conduction velocities are not slowed to the degree expected in CMT1. Also, the clinical features of POEMS suggest an acquired disorder with a rapidly progressive polyradiculoneuropathy in contrast with a lifelong progressive length dependent neuropathy typical of CMT1.16 Almost all of the different demyelinating components (conduction velocities, distal latencies, F wave latencies) were the same or slower than in CIDP. This in addition to the lack of conduction block and temporal dispersion suggests more uniform demyelination along the nerve rather than focal involvement of selected nerve fibres or segments.

Needle EMG of POEMS patients demonstrated a greater degree of fibrillation potentials in both distal predominant limb muscles. This finding may speak to not only the degree of axonal loss, which is more significant than in CIDP controls, but to the prominent relentless progression of the peripheral neuropathy in POEMS patients. We wondered if POEMS patients were evaluated later into the course of their illness than CIDP, but think this is less likely as Neuropathy Impairment Scores were similar. Perhaps the explanation is that the deficits in POEMS syndrome result mainly from axonal loss in contrast with CIDP where the deficits mainly result from demyelination.

The presence of proximal demyelinating features and distal axonal loss on NCS and needle EMG is supported by the pathological features reported in POEMS syndrome.17 However, the exact cause of the pathological changes is unknown. Considerations include: (1) an immune mechanism, due to M protein mediated activation of the complement dependent pathway leading to disruption of the blood–nerve barrier and (2) possible effect of increased VEGF levels leading to small vessel permeability and endoneurial oedema.18 Pathological specimens have shown an increase in numbers of small epineurial vessels but it is unclear what effect if any these vessels have on nerve function.19

Clinically, TLI was most informative on nerve conduction tests in discriminating cases of POEMS from CIDP. In our population, the median motor TLI had an acceptably high sensitivity and specificity, suggesting that this test may be applicable in distinguishing these disorders electrophysiologically. The median TLI value in POEMS syndrome was similar to that reported previously.14 15 TLI has been shown to be smaller in IgM MGUS neuropathy patients than in CIDP patients.20 The question could arise whether TLI would be unable to distinguish POEMS and MGUS neuropathy patients. We think this potential problem is unlikely as the TLI in POEMS is actually larger than in CIDP patients.

In summary, NCS are helpful in distinguishing POEMS syndrome from CIDP. Patients with POEMS syndrome demonstrate uniform demyelination (slowed conduction velocities, long distal latencies and long F wave latencies) with infrequent conduction block and temporal dispersion, and lack of sural sparing. There is preferential involvement of the intermediate nerve segment in POEMS compared with CIDP. There is greater axonal loss which results in clinical deficits in POEMS patients. These findings imply that the pathology of POEMS syndrome is diffusely located along the nerve whereas the pathology of CIDP is multifocal and probably predominantly proximal and distal. We acknowledge that in individual cases these features may not be able to distinguish a POEMS patient from a CIDP patient. However, the findings of uniform demyelination in an acquired neuropathy should prompt clinicians to consider the diagnosis of POEMS syndrome and look for other features, including organomegaly, endocrinopathy, monoclonal protein and skin changes. All patients with a sensorimotor polyradiculoneuropathy with demyelinating features on electrodiagnostic testing (suggestive of CIDP) should have serum and urine protein electrophoresis and immunofixation, for evaluation of a monoclonal protein, and skeletal bone survey.


The authors thank Jenny Davies for her assistance in calculating nerve conduction percentile values.


View Abstract


  • See Editorial commentary, p 474

  • Linked article 301706.

  • Competing interests None.

  • Ethics approval The study was approved by the institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles