Classifications and treatment responses in chronic immune-mediated demyelinating polyneuropathy

Patients.

We retrospectively collected 158 patients with a suspected diagnosis of CIP of autoimmune origin. All patients were examined at the authors’ neuroimmunology clinics during the 5-year period from 1998 to 2003. Because of lacking a gold standard for the diagnosis of CIDP as well as demyelinating inflammatory polyneuropathies as a whole, we used pragmatic inclusion criteria published elswhere.^19,34 To report a wide spectrum of CIP, we here also included MMN as a peripheral motor syndrome, diagnosed separately according to the American Association of Electrodiagnostic Medicine criteria.³⁵ One hundred two patients were included in the study, because 1) they had symptoms and signs of a progressive, relapsing or monophasic, symmetric, or asymmetric (multifocal) acquired peripheral neuropathy of 2) at least 2 months in duration and 3) electrophysiologic evidence for demyelination according to published and used criteria (see below).^19,34,35 Patients with evidence for hereditary, toxic, metabolic, infectious, amyloid, or paraneoplastic neuropathy and patients with Guillain–Barré syndrome, critical care neuropathy, or other neuromuscular diseases or malignancy were excluded. Because vasculitis usually results in axonal neuropathy, we excluded 5 patients. If monoclonal immunoglobulin was present, a hematologic malignancy was excluded. Patients with a marked bilateral difference of upper or lower extremities in strength (two or more Medical Research Council grades) or sensory function were defined as having an asymmetric form of CIP. Figure 1 shows an overview of the study design.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Study flowchart and analysis procedure

We performed two different analyses with respect to clinical course (Analysis A) and clinical pattern (Analysis B).^{1 a}Computed multicenter search for the working diagnosis of chronic immune-mediated demyelinating polyneuropathy (CIP). ^bPatients were excluded because they had a diagnosis of polyneuropathy of other cause (n = 41; 22 were negative for demyelination criteria,^19,34 5 had vasculitis, 4 had plasmocytoma, 3 had Guillain–Barré syndrome, 2 had critical care neuropathy, 1 each had CREST [calcinosis, Raynaud phenomenon, esophageal motility disorders, sclerodactyly, and telangiectasia] syndrome, hereditary motor and sensory neuropathy, borreliosis, neuromyotonia, or AIDS), incomplete clinical data (n = 6), or evidence for primarily spinal pathology (n = 7). ^cPatients fulfilling inclusion criteria outlined above. ^dMonophasic course: one disabling attack of CIP of at least 8 weeks. ^eRelapsing–remitting course: at least two disabling attacks with improvement for more than 1 point on the Rankin scale and stable course between attacks. ^fChronic–progressive course: progressive disability without relapse. ^gAccording to reference ¹. ^hAccording to references ^{35 and 37}. ⁱAccording to reference ⁴⁵. ^jAmong the 24 patients with monoclonal gammopathy of uncertain significance (MGUS), their phenotypes were chronic inflammatory demyelinating polyneuropathy (CIDP)–like (n = 14), distal acquired demyelinating sensory polyneuropathy (DADS)–like (n = 7), and multifocal acquired demyelinating sensory and motor polyneuropathy (MADSAM)–like (n = 3); none of the patients with MGUS had a multifocal motor neuropathy (MMN)–like phenotype.

Classifications, electrophysiology, and CSF analysis.

Symmetric CIP syndromes were classified using the original AAN research criteria from 1991 for (symmetric) CIDP,¹⁰ the modified AAN criteria from 2001,¹ and the recently published criteria by the EFNS/PNS.¹⁸ Patients’ disability was measured by a modified Rankin scale³⁶: the clinical course was classified as monophasic (MO), relapsing–remitting (RR) with or without complete remission, or chronic–progressive (CP).

All centers performed standardized motor and sensory nerve conduction studies on ulnar, median, radial, peroneal, tibial, and sural nerves depending on the clinical manifestation according to the recommendations of the International Federation of Clinical Neurophysiology (DGKN/IFCN).³⁷ Except for MMN, demyelination was diagnosed according to published criteria,^19,34 when three of the following four criteria were met:

1. conduction velocity of less than 90% of the lower limit of normal (LLN) (i.e., <45 m/second in the arms and 36 m/second in the legs), if the amplitude exceeds 50% of the LLN; less than 80% of the LLN (40 m/second in the arms and 32 m/second in the legs), if the amplitude is less than 50% of the LLN

2. distal latency exceeding 115% of the upper limit of normal (ULN), if the amplitude is normal (4.8 ms in the median nerve and 5.9 ms in the legs); exceeding 125%, if the amplitude is less than the LLN (5.3 ms in the median nerve and 6.4 ms in the legs)

3. proximal or distal amplitude ratio of less than 0.7

4. F-wave latency exceeding 125% of the ULN (36 ms in median, 38 ms in ulnar, and 63 ms in tibial and peroneal nerves) or absent F-waves in one or more nerves

CSF samples were analyzed for cell count, total protein (normal up to 420 mg/L), IgG, IgM, and IgA. Oligoclonal bands were analyzed by isoelectric focusing. Sural nerve biopsies were performed in 40 patients. None of the biopsies revealed vasculitis, amyloid deposits, or other specific pathology.

A positive treatment response was defined as improvement of at least one point on the modified Rankin scale for at least 1 week. To attribute the improvement to the immunomodulating therapy, the amelioration was requested to be present within 2 months after starting steroids, 2 weeks after performing IVIg or plasmapheresis treatment, or 4 months after beginning cyclophosphamide.

Biostatistical analysis.

For statistical evaluation, we calculated the sensitivity (TP rate, i.e., the probability that the test result is positive among CIP patients); specificity (TN rate, i.e., the probability that the test result is negative among non-CIP patients); positive predictive value (PPV, i.e., the probability that a patient has CIP when restricted to those patients who fulfil the respective criterion); negative predictive value (NPV, i.e., the probability that a patient will not have CIP when restricted to all patients who do not fulfil the respective criterion); and finally, likelihood ratio (LHR, i.e., how much the odds of having CIP or a positive treatment response increase when the respective criterion is positive). Significance was calculated two-sided at the 5% level, using the nonparametric Mann–Whitney test for unlinked samples (differences of medians) and the nonparametric χ² and Fisher exact tests (categorical differences). Testing 21 statistical hypotheses, the Bonferroni correction for multiple testing was 0.05/21 = 0.0024. Consequently, p < 0.0024 was considered significant. SPSS 11.0 software was used for statistical calculation.

Clinical profiles.

The clinical characteristics of the 102 included patients with CIP are summarized in table 1. The majority (74%) showed a steady chronic progression (CP) of symptoms, whereas an RR or MO course was observed in 20% or rather 7%. In these latter two groups, a symmetric neurologic deficit, consistent with the classic CIDP concept, was observed almost exclusively. In contrast, in CP patients, asymmetric deficits were found in 30% of the patients with a significant proportion of purely motor (19%) or sensory (15%) involvement. A predominantly distal maximum of the neurologic deficit was observed in all patient groups.

A male preponderance was present in all groups, with an overall male-to-female ratio of 4.4. CP-CIP patients had the lowest disability at diagnosis compared with the RR or MO patients (not significant). In contrast, patients with MO-CIP showed the highest initial disability (table 1) but were by far the best responders to IVIg (table 4). CSF protein was increased in 75%, and only 14% showed mild pleocytosis (13 patients had 5 to 13 cells per μl). MGUS was diagnosed in patients from all three groups, remarkably, without significant difference in frequency.

Diagnostic classifications.

To evaluate the validity of three available diagnostic classifications for CIDP, all patients with a symmetric distribution of their neurologic deficit (proximal or distal; motor, sensory, or sensorimotor; with or without monoclonal MGUS; n = 76) were grouped into four categories under the terms of the respective classification nomenclature^1,10,18 (table 2). Though the EFNS/PNS criteria include MADSAM as atypical CIDP, such patients were not included here, because the original and modified AAN criteria only focus on symmetric CIP. According to the original AAN criteria, 40 patients (53%) were grouped into “definite” or “probable” CIDP, whereas 63 (83%) and 72 (95%) fit into the “definite” or “probable” categories of the other two.

The EFNS/PNS and modified AAN criteria showed a considerably higher sensitivity but no relevant difference of specificity (table 2). The positive predictive values were similar in all classifications but highest using EFNS/PNS (0.97). The higher negative predictive values of the EFNS/PNS and the modified AAN criteria, however, suggest their stronger practical differential diagnostic impact. The relevance of electrodiagnostic and CSF analysis in confirming the diagnosis is similar in all classifications (table 3), conduction studies being the most important feature, especially using the original AAN criteria (LHR 9.0).

CSF analysis is a mainstay in the diagnosis of polyneuropathy. However, it has rarely been quantified. Classic CIDP has the highest CSF protein at time of diagnosis, whereas MMN has the lowest (figure 2). Basically, CSF analysis does not play an important role in the diagnostic classifications of CIDP (LHR 1.4 vs 1.2 vs 1.2), notwithstanding their high sensitivity of 0.88 to 1.0 (table 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 CSF protein level in the differential diagnosis of chronic immune-mediated demyelinating polyneuropathy

Patients with monoclonal gammopathy of uncertain significance (MGUS) are analyzed separately. Shown are medians, 25th and 75th percentiles, and outliers. The single highest CSF protein level was found in chronic inflammatory demyelinating polyneuropathy (CIDP; 2,500 mg/L). Median CSF protein in CIDP (997 ± 476 mg/L) is higher than in distal acquired demyelinating sensory polyneuropathy (DADS; 812 ± 477 mg/L; p < 0.03) and multifocal motor neuropathy (MMN; 521 ± 179 mg/L; p < 0.0001). The CSF protein medians in patients with multifocal acquired demyelinating sensory and motor polyneuropathy (MADSAM) is 794 ± 254 mg/L, and that in patients with MGUS is 761 ± 408 mg/L. The median in DADS is higher than in MMN (p < 0.04). When MGUS patients are divided into their phenotype, their CSF protein levels are 870 ± 460 mg/L in CIDP-like, 703 ± 347 mg/L in DADS-like, and 460 ± 40 mg/L in MADSAM-like patients.

Treatment response.

As first-line treatments, IVIg (n = 76) or corticosteroids (n = 14) were used. As second- or third-line treatments, corticosteroids (n = 29), azathioprine (n = 12), cyclophosphamide (n = 9), plasmapheresis (n = 5), interferon beta (n = 2), mycophenolate mofetil (n = 1), or methotrexate (n = 1) was used.

Treatment response to IVIg varied significantly according to the clinical course of the neuropathy (table 4). This is also reflected by different TP rates (table 5). Seventeen percent of the study population had a complete clinical remission with or without treatment (table 4). After stratifying CIP into CIDP, DADS, MADSAM, and MMN (MGUS not included),¹ DADS neuropathy showed the best prognosis and the highest remission rate (42%; table 4). Eighty-two percent of the IVIg-treated patients improved by 1 or more Rankin points. Treatment responses to IVIg or corticosteroids did not differ significantly among subgroups (table 4). Corticosteroid first-line monotherapy led to improvement in 57%, and plasmapheresis led to improvement in 71%. Six of nine patients treated with cyclophosphamide (2 of 3 classic CIDP, 1 of 1 MADSAM, 1 of 3 MMN, 2 of 2 MGUS) showed a clear clinical benefit.

Analysis of response markers (table 5) revealed that all classifications are similar according to PPV. The NPV was the highest using the EFNS/PNS criteria (0.33). From this it can be derived that the number needed to treat is low with all classifications (1.7 for AAN/modified AAN and 1.6 for EFNS/PNS) for CIDP. The highest TP rate for treatment response was achieved using the EFNS/PNS criteria (0.96) for diagnosis, whereas the highest TN rate appeared after using the AAN criteria (0.45). Remarkably, patients fulfilling the modified AAN criteria for CIDP and also having more than twofold elevated CSF protein (≥840 mg/L) are 5.5 times more likely to respond to IVIg than for those with a lower CSF protein (3.3 using the AAN criteria; 2.5 using the EFNS/PNS criteria).