Article Text
PDF
Abstract
Background The diagnostic value of new criteria of the European Academy of Neurology/Peripheral Nerve Society (EAN/PNS) for chronic inflammatory demyelinating polyneuropathy (CIDP) is unknown.
Methods We performed a retrospective study of fulfilment of EAN/PNS 2021 criteria on 120 consecutive patients with a clinical diagnosis of ‘suspected CIDP’ and objective treatment response, attending University Hospitals Birmingham, UK. Specificity was evaluated versus 100 consecutive controls.
Results The sensitivity of EAN/PNS criteria for ‘CIDP’ was 83.3%. The sensitivity for ‘CIDP’ or ‘possible CIDP’ was 93.3%. Specificity was of 94% for ‘CIDP’ and 79% for ‘CIDP’ or ‘possible CIDP’. No sensitivity/specificity differences were ascertained with previous versions (‘CIDP’: sensitivity: 83.3% vs 81.3%, p=0.74, specificity: 94% vs 96.1%, p=0.38, respectively; ‘CIDP’ or ‘possible CIDP’: sensitivity: 93.3% vs 96.7%, p=0.25 and specificity: 79% vs 69.2 %, p=0.09, respectively). F-wave prolongation, proximal and distal temporal dispersion were the most likely parameters to contribute to false positives, whereas distal motor latency was the least likely. No impact of sensory electrophysiology could be ascertained. ‘Typical CIDP’ represented 79% of the CIDP cohort. The largest component of the ‘variant CIDP’ group was represented by focal/multifocal forms (14%). With new criteria, 6.7% of the cohort did not meet requirements, among whom the majority (75%) had paranodopathy or chronic immune sensory polyradiculopathy (CISP).
Discussion The sensitivity and specificity of new EAN/PNS criteria for CIDP is equivalent to that of previous versions. The exclusion of paranodopathies and CISP from the CIDP spectrum impacts on management of a non-negligible proportion of treatment-responsive patients.
- CLINICAL NEUROLOGY
- NEUROPHYSIOL, CLINICAL
- NEUROPATHY
- EMG
- NEUROMUSCULAR
Data availability statement
Data are available on reasonable request. Available on reasonable request.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
The diagnostic reliability of previous versions of the European Federation of Neurological Societies/Peripheral Nerve society (EFNS/PNS) have been shown by previous validation studies.
WHAT THIS STUDY ADDS
The current study investigates the diagnostic value of the updated 2021 version of the EAN/PNS criteria and adds important novel data on the new criteria.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
The findings of this study demonstrate the value of the newer version of the criteria for clinical practice and research, while raising important questions regarding inclusion of rarer but related syndromes within the CIDP spectrum, for a non-negligible minority of affected patients.
Introduction
Chronic inflammatory demyelinating polyneuropathy (CIDP) is the most common dysimmune neuropathy and commonly affects patients during active phase of life.1 Clinical presentations of CIDP are heterogeneous, with possible different pathophysiological differences and treatment responses, but persistent absence of a biomarker in vast the majority of affected subjects.2 Timely diagnosis of CIDP is essential as it may lead to considerable functional deficits, including pain and fatigue, especially when appropriate treatment strategies are delayed.3 4 Several sets of diagnostic criteria have been published with different combinations of the required clinical and electrodiagnostic features and variable diagnostic accuracy.5–9 Consequently, misdiagnosis is not uncommon in routine clinical practice.10 11 The joint guideline on CIDP from the European Academy of Neurology (EAN) and Peripheral Nerve Society (PNS) is the most widely used set of consensus criteria.12 The recent 2021 update of the previous versions, published in 2005 and 2010, aimed to address several limitations including simplification of diagnostic categories, revision of supportive and exclusion criteria.13–15 The criteria revision was hence to optimise diagnostic accuracy and improve patient outcomes. Electrodiagnostic support is still the main diagnostic requirement for CIDP in this EAN/PNS guideline. Compared with the previous versions, the main differences with regards to the electrophysiological component of the updated diagnostic criteria are (1) the combination of previously defined ‘definite CIDP’ and ‘probable CIDP’, into a single category of ‘CIDP’, with motor conduction block now receiving the single definition of a reduction of at least 30% of the compound muscle action potential (CMAP) amplitude (2) the use of filter-dependent cut-off values defining distal CMAP dispersion and (3) the formal inclusion of sensory nerve conduction data within the criteria.13 14
Previous validation studies have demonstrated high sensitivity and specificity of the previous versions of the criteria.6 9 16–18 However, the potential impact on diagnostic reliability, of the above-mentioned modifications is currently unknown. We aimed to evaluate the sensitivity and specificity of the updated version of the criteria in a population with a clinical diagnosis of CIDP and objective response on appropriate immunomodulatory therapy. We intended to compare the results with those obtained with previous versions and to determine the reasons for eventual differences and findings that may affect clinical and research practice.
Methods
Patients
We performed a retrospective review of the electronic records of all consecutive patients presenting with a clinical diagnosis of ‘suspected CIDP’, as per initial clinical evaluation in our outpatient clinic, attending the Inflammatory Neuropathy Service, at Queen Elizabeth Hospital, University Hospitals Birmingham, between July 2014 and February 2022. We defined ‘suspected CIDP’ as subjects that fulfilled the required clinical features of CIDP including the typical form, or of any clinical variant except distal CIDP.13 14 19 Distal CIDP, defined within the guidelines as ‘presenting (clinically) with distal upper and lower limb sensory loss and gait instability’ was excluded from the original review because this definition may be met in various forms of distal non-CIDP neuropathies. The inclusion of all such clinical cases may, therefore, obviously lead to erroneously low sensitivity calculation for the disease overall, particularly when not considering immunotherapy response as mandatory. However, cases subsequently identified as distal CIDP, through electrophysiological evaluation and meeting criteria specified in updated EAN/PNS Guidelines, were added to the cohort.13 14 We selected patients who had received appropriate treatment for the diagnosis of CIDP and for whom documented evidence of response was available in our records. We reviewed detailed clinical and electrophysiological records in each ‘suspected CIDP’ patient with objective treatment responsive. In addition, we evaluated 100 consecutive neuropathy controls, seen by our service during the same period, with a diagnosis of ‘non-CIDP neuropathy’, made on exclusively on clinical grounds, as documented in available records, at initial clinical assessment. A ‘non-CIDP’ neuropathy was defined for the purposes of the current analysis as a sensory, motor, or mixed sensory and motor neuropathy (1) involving at least one limb, (2) without bilateral symmetrical proximal weakness and (3) without proprioceptive sensory loss. Subjects with Charcot-Marie-Tooth disease type 1, amyloid polyneuropathy, POEMS (polyneuropathy, organomegaly, endocrinopathy, M-protein, skin) syndrome and with monoclonal IgM gammopathy associated anti-MAG (myelin associated glycoprotein) neuropathy were excluded from the control group.
Treatment response
We defined treatment response as amelioration by ≥1 point on the Overall Neuropathy Limitation Scale within 12 weeks of commencing any appropriate therapy.20 21 In patients not meeting this requirement, response we used the following criteria for objective improvement on treatment: increase of ≥4 raw points on the Inflammatory Rasch-Built Overall Disability scale and/or of ≥5 kg on Jamar grip strength of the dominant or more affected hand.20–22
These definitions were decided in view of their long-term use in our unit as well as previous minimum important difference cut-offs, in our cohort of patients, prior to publication of the guideline update.20 21
Electrodiagnostic studies
Median, ulnar, common peroneal and tibial motor nerves as well as median, ulnar, radial and sural sensory nerves were studied using standard equipment and procedures as described previously (figure 1),23 by specialist medical staff trained and qualified in electrophysiology. Minimum electrophysiological study requirements were of four tested motor nerves and three tested sensory nerves. All available data were analysed using locally-derived normative values and available recommendations for each electrophysiological parameter (figure 1), as per the updated EAN/PNS Guidelines.13 14 Only motor nerve conduction data of unilateral forearm and forelimb segment studied were considered. We also evaluated the value of sensory conduction criteria.13 14
Summary of standardised electrodiagnostic protocol. Annotation: overview standardised sites of stimulation and recording in routine Nerve Conduction Study (NCS) protocol, evaluating for presence of electrophysiological features compatible with demyelination (A) and fulfilment of sensory conduction criteria (B). limb temperature is maintained at 32°C and 30°C, at wrist and ankle, respectively. Sensory nerve action potential (SNAP) amplitudes are evaluated from peak-base. Compound muscle action potential (CMAP) duration is evaluated as interval between onset first negative peak and return to baseline of the last negative peak, using lower frequency filter setting of 2 Hz. CB, conduction block (updated guideline no longer distinguishes between definite and probable CB, and is defined on reduction of the proximal relative to distal negative peak CMAP amplitude, without correction for concurrent temporal dispersion over same segment); DML, distal motor latency; G0, ground electrode; MCV, motor conduction velocity; R1, recording electrode; R2, reference electrode; S1, distal stimulation site; S2, proximal stimulation site; SCV, sensory conduction velocity; TD, temporal dispersion.
Fulfilment of electrophysiological requirements for CIDP in the neuropathic controls was established, using, as in the ‘suspected CIDP’ cohort, exclusively four motor nerves studied unilaterally in forearm and foreleg segments. Fulfilment of sensory conduction criteria was determined. Hence, we evaluated the specificity of the two levels of diagnostic certainty for CIDP, using the updated EAN/PNS criteria in this control cohort.13 14
We analysed fulfilment of clinical and electrophysiological criteria for CIDP as per updated guidelines but did not consider, in the current analysis, other supportive tests such as cerebrospinal fluid (CSF) analysis and imaging. These tests are not recommended to be used routinely by the updated guidelines for subjects meeting electrophysiological requirements for CIDP, but only for those fulfilling requirements for possible CIDP. Furthermore, they may pose difficulty with uncertainty regarding accurate normal range values and are of lower specificity. Therefore, we only included cases with objective treatment response, as alternate reference standard, which also represents a supportive feature of CIDP.
Statistical analysis
Diagnostic sensitivity of the updated criteria was determined in this cohort of patients with treatment responsive ‘suspected CIDP’.13 14 The precise diagnostic category, including level of diagnostic certainty and subtype was established for each patient. Sensitivity and specificity of the updated criteria were evaluated and compared with those of previous versions.12 We used t-tests and comparison of proportions, by Fisher’s exact tests to comparison means. We used SPSS software (IBM, V.26) for statistical analysis.
Results
We identified, of a total of 156 subjects with ‘suspected CIDP’, 120 subjects with documented evidence of treatment response as defined above. Patients’ and controls’ characteristics are summarised in table 1. Figure 2 summarises the distribution of different patient subgroups with regard to criteria fulfilment and disease subtype.
Summary of characteristics of patients and controls
Summary distribution patient groups and criteria fulfilment. Annotation: the previous versions of EAN/PNS guidelines defined the current set of distinct variants of CIDP, CISP and paranodopathies all as single group of ‘atypical’ CIDP. The present splitting results in changes in sensitivity, and specificity. Furthermore, extensive description of treatment strategies for CISP and paranodopathies is not part of current CIDP guideline. CIDP, chronic inflammatory demyelinating polyneuropathy; CISP, chronic immune sensory polyradiculopathy; EAN/PNS, European Academy of Neurology/Peripheral Nerve Society.
Diagnostic classification updated EAN/PNS criteria
The distribution of patient numbers as per criteria fulfilment and subtype is summarised in figure 2. Of the 120 patients, 100 (83.3%), met updated EAN/PNS criteria for a diagnosis of ‘CIDP’. Twelve (10%) met criteria for ‘possible CIDP’. The sensitivity of the criteria in this cohort, for a diagnosis of ‘CIDP’ OR ‘possible CIDP’ was hence of 93.3%. Eight subjects (6.7%) did not meet criteria for CIDP. Of those meeting criteria for ‘CIDP’, 79 (79%) had typical CIDP and 21 (21%) had a CIDP variant. The variant group comprised 14 subjects (14%) with focal/multifocal CIDP, 4 (4%) with motor CIDP, 1 (1%) with motor-predominant CIDP, 1 (1%) with sensory-predominant CIDP and 1 (1%) with distal CIDP. Of the 12 subjects with ‘possible CIDP’, 6 had a typical form and 6 had a variant form, of which 3 had focal/multifocal disease, 1 had sensory disease and 2 had distal disease. Of the eight subjects who did not meet the updated criteria, one had a typical form, one had a motor-predominant presentation, one had chronic immune sensory polyradiculopathy (CISP) and five had a paranodopathy (with anti-NF155 antibodies in four and anti-contactin1 antibodies in one).
Diagnostic accuracy updated EAN/PNS criteria
We found that the sensitivity of the updated criteria in our cohort for typical ‘CIDP’ was 91.9% (79/86) and that for typical ‘CIDP’ or ‘possible CIDP’ was 98.8% (85/86). Sensitivity for variant ‘CIDP’, including only focal/multifocal, disease was 82.4% (14/17) and that of ‘CIDP’ or ‘possible CIDP’ was 100% (17/17). Combined sensitivity for typical and focal/multifocal ‘CIDP’ was hence 93.9% (93/99) and was of 99% (99/100) for typical and focal/multifocal ‘CIDP’ or ‘possible CIDP’. Hence, and of interest, the sensitivity of the criteria, combining ‘CIDP’ and ‘possible CIDP’ subcategories, was greater in typical and focal/multifocal disease versus that in the whole cohort (99/100 vs 112/120; p=0.042).
The control population consisted clinically, of 62 subjects with a distal sensory neuropathy, 1 with a multifocal sensory neuropathy, 35 with a distal sensory and motor neuropathy, 1 with a lumbosacral radiculoplexus neuropathy and 1 with a distal motor neuropathy. Twenty-four patients had Charcot-Marie-Tooth disease type 2. Six controls met updated EAN/PNS criteria for ‘CIDP’ and 15 met criteria for ‘possible CIDP’. The specificity versus controls was hence of 94% for the ‘CIDP’ level of diagnostic certainty and 79% for the combined ‘CIDP’ or ‘possible CIDP’ level of diagnostic certainty. Electrodiagnostic abnormalities leading to false positives were found in 10/21 subjects for the common peroneal nerve, in 9/21 for the tibial nerve, in 6/21 for the median nerve and in 4/21 for the ulnar nerve. With regard to individual parameters, F-wave abnormalities were present in 7/21, proximal temporal dispersion in 6/21, distal CMAP prolongation in 5/21, conduction block in 4/21, reduced motor conduction velocity in 3/21 and prolonged distal motor latency in 0/21. Hence, no nerve was significantly more likely to produce false positives. On the other hand, F-wave prolongation, proximal temporal dispersion and distal CMAP prolongation were all more likely to contribute to false positives than distal motor latency prolongation (7/21 vs 0/21, p=0.009; 6/21 vs 0/21, p=0.02; 5/21 vs 0/21; p=0.048, respectively).
We analysed fulfilment of the sensory conduction criteria, as detailed by the updated EAN/PNS Guidelines in both the CIDP cohort and controls. In the CIDP group, 112 subjects (93.3%) met sensory criteria in 2 nerves, 1 (0.8%) met criteria for a single nerve and 7 did not meet the sensory criteria. In the control group, 60 subjects (60%) met sensory criteria in 2 nerves, 8 (8%) met the sensory criteria in a single nerve and 32 (32%) did not meet the sensory criteria. Hence, the sensitivity of fulfilment of sensory criteria in two nerves was greater than that of updated motor criteria for ‘CIDP’, met, respectively, by 93.3% vs 83.3% of the cohort (p=0.026). No sensitivity difference was however ascertained in comparison to that of updated motor criteria for ‘CIDP’ or ‘possible CIDP’ (93.3% vs 93.3%; p=1).
Comparison yield updated EAN/PNS criteria with older versions
In comparison to the validation study performed for the first version of the criteria, no differences of sensitivity or specificity could be ascertained for the highest level of diagnostic certainty, that is, previously defined as ‘definite’ or ‘probable CIDP’, now defined as ‘CIDP’ (sensitivity: 83.3% vs. 81.3%, p=0.74 and specificity: 94% vs 96.1%, p=0.38, respectively). Similarly, no differences were found with the lowest level of diagnostic certainty, that is, ‘possible CIDP’ (sensitivity: 93.3% vs. 96.7%, p=0.25 and specificity: 79% vs 69.2 %, p=0.09, respectively).
Discussion
The revision of the guidelines aimed to improve diagnosis and patient outcomes. Our study, which focused on the diagnostic aspect, showed that the sensitivity and specificity of the updated version of the EAN/PNS criteria for clinically ‘suspected CIDP’ are similar to those of previous versions. In contrast, our study did not show obvious, useful added diagnostic value of the sensory conduction criteria which are now a mandatory part of the updated version. We also found that most patients not meeting criteria had disease forms now excluded from the initial CIDP spectrum. Nevertheless, the simplified diagnostic categories and inclusion of specific variants of CIDP are an improvement for its use in clinical practice.
Our results are indicative of the excellent diagnostic value of the criteria, well-demonstrated through comparative studies with other sets of criteria, previously.6 9 16–18 Consequently, omitting terminology no longer considered appropriate, such as ‘atypical’ and ‘definite/probable’ CIDP, and simplifying diagnostic categories has not impacted the detection rates of patients with suspected CIDP that respond to treatment. However, we could not find an improvement in reliability in our study population. It is possible that the similar sensitivity of the updated, compared with that of older versions of the criteria, despite exclusion of paranodopathies and CISP, is due to the sample size studied, identical to that of a previous multicentre validation study (120 vs 123 subjects). Inclusion of excluded cases of CISP and paranodopathy would have resulted in improved diagnostic sensitivity for ‘CIDP’ in this cohort by 5%, from 83.3% to 88.3%. However, this would be still similar to the sensitivity of older criteria (88.3% vs 81.3%; p=0.15).6 On a clinical management basis, the exclusion of 5% of our ‘suspected CIDP’ cohort, from the CIDP spectrum, would have represented a practical disadvantage and unhelpful complication, due to absence of widespread antiparanodal antibody testing at referring centres as well as with regards to requirements for approval of immunoglobulin therapy in the UK, effective in CISP as well as in a proportion of paranodopathies.24 25
With regard to specificity, the results obtained are clearly directly related to the control population studied. We do not believe the demographic differences affected our results, which were analysed using locally derived normative values from adults of variable age and gender, consistent with published values from other laboratories. On the other hand, our clinical criteria for selection of controls, which was one of an initial clinical diagnosis of ‘non-CIDP’ neuropathy, impacted on the specificity calculations, as included mostly subjects with pure distal small fibre presentations. Alternative methods would however also have posed other significant problems. Inclusion of other demyelinating neuropathies, genetic or acquired, would have artificially caused decline of specificity, as the criteria studied were not devised to differentiate between different demyelinating neuropathies. Hence, the issue of evaluation of specificity of electrodiagnostic criteria for CIDP remains difficult to tackle, particularly given the absence of a diagnostic gold-standard and the heterogeneity of the variant forms. In this regard, we added, after selection of clinically selected cases, those with electrophysiology consistent with CIDP but who had a length-dependent presentation and who therefore met requirements for the distal CIDP variant.13 14 This illustrates the difficulty in ascertaining the value of the criteria and also, their limitations, as distal CIDP cannot be suspected on clinical grounds without electrophysiology, as opposed to all other forms of the disorder. Inclusion of all distal neuropathy presentations would of course have resulted in extremely and artificially low sensitivity of the criteria, rendering them of no practical value. With regards to specificity, we found that no nerve was more likely to show abnormalities resulting in false positives whereas distal latency prolongation was significantly less likely to produce misclassification than F-wave prolongation, proximal temporal dispersion and distal CMAP prolongation, particularly in presence of predominant axonal disease as was the case in our control population. This is of importance, as caution with isolated abnormalities of the latter three parameters appears therefore appropriate. It is also finally noteworthy that consideration of the different CIDP subtypes indicates that the criteria are particularly useful in typical presentations and focal/multifocal disease. This suggests clinical splitting of disease forms may be more appropriate than when done on the basis of presence of specific antibodies or presence of electrophysiological features of demyelination, as is currently the case.26–28
At present, a reliable diagnostic biomarker is lacking, and an alternative diagnostic gold-standard remains absent. Therefore, we chose to only include suspected CIDP patients who had demonstrated objective response to immunotherapy as alternate reference standard. Although refractory CIDP clearly exists, other diagnoses may be more likely in case of refractoriness, justifying further investigations in routine clinical practice.11 Therefore, despite likely exclusion of some cases of genuine CIDP through this methodology, we believe we have reduced the possible bias of inappropriate inclusion of other disorders. Other limitations include retrospective design and smaller sample sizes of the CIDP variants. However, most CIDP variants other than multifocal CIDP are rare and inclusion of large numbers of the CIDP variants likely leads to selection bias and overfitting as these are very uncommon in routine clinical practice. Hence, we did not use nerve imaging (qualitative MR neurography, performed in our unit, has poor interobserver reliability29 and nerve ultrasonography is not routinely performed at our institution) and CSF testing (only considered in appropriately selected cases in our practice) as supportive tests in this analysis. Their contribution remains to be further determined but likely smaller than in previous versions of guideline. The exclusive focus on clinical, electrophysiological and treatment response aspects represents a limitation of our study. It may be useful in future to study the added value of these tests with current guidelines.
In conclusion, updated diagnostic CIDP criteria do not appear to be limited by reduction of diagnostic categories nor are they more sensitive or specific than older versions. However, some treatment responsive groups are no longer considered as part of the CIDP spectrum, which, importantly, could impact on their management as reimbursement may be limited. Further thought and discussion about optimal disease subtype classification and exclusion criteria is therefore warranted.
Data availability statement
Data are available on reasonable request. Available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was reviewed, approved and registered as a Clinical Audit of clinical practice, by the Clinical Audit Office (CARMS no. 15366), University Hospitals Birmingham NGS Foundation Trust, UK.
References
Footnotes
Contributors YAR: data analysis, data interpretation, drafting of first version of manuscript. SA: data collection, data analysis, review of manuscript for intellectual content. LKL: data collection, review of manuscript for intellectual content HSG: data interpretation, drafting of first version of manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests LKL and SA have no disclosures.YAR has received speaker/consultancy honoraria from LFB, Polyneuron and Argenx and has received educational sponsorships from LFB and CSL Behring and has obtained research grants from LFB and CSL Behring. HSG has received research grants from Prinses Beatrix Spierfonds and travel grants speaker fee from Shire/Baxalta and Takeda.
Provenance and peer review Not commissioned; externally peer reviewed.