Marieke van Rosmalen

General discussion 131 8 and recommendations in this chapter on these criteria for CIDP. For MMN, such comparative studies of sets of diagnostic criteria do not exist and we will use the Utrecht criteria for further evaluation as these criteria also predict treatment response to immunoglobulins. 10,22–24 The challenge of diagnosing CIDP or MMN relies mainly on the fact that the diagnostic tools used in both sets of criteria have their limitations in current clinical practice. In both sets of criteria abnormalities on NCS are the key feature of diagnosis of CIDP and MMN. However, NCS does not always show the required specific electrophysiological features. Signs of demyelination could be easily missed as the NCS protocols are extensive and require specific expertise. If NCS does not fulfill the electrodiagnostic criteria, diagnosis remains uncertain. This led in the nineties to the search for additional tools that support diagnosis of CIDP or MMN. Early studies on magnetic resonance imaging (MRI) of the cauda equina and the nerve roots of the lumbosacral and brachial plexus showed thickening, a hyperintense signal on T2-weighted imaging and enhancement of the nerve roots after infusion with gadolinium. 25–29 This led to the inclusion MRI as a supportive criterion in current guidelines for diagnosis of CIDP and MMN. Unfortunately, the diagnostic value of MRI has not been studied in detail previously. MRI is qualitatively assessed by (neuro)radiologists which is potentially a major limitation in current clinical practice. Table 8.1 summarizes the diagnostic criteria for CIDP and MMN and their limitations. Qualitative assessments of brachial (or lumbosacral) plexus MRI are subjective and large systemic studies that report objective cut-off values of nerve size are lacking. This may lead to large interrater variations as the reliability of the current assessment method has not been systematically studied. In chapter 3 we therefore performed a study that evaluated interrater reliability of the current qualitative assessment approach of brachial plexus MRI. We found that raters agreed in only 52% of all brachial plexus images, with a Cohen’s kappa indicating minimal agreement and a poor reliability (0.30, 95% confidence interval 0.14 – 0.46). Difficulties in assessing the brachial plexus were mostly related to distinguishing more subtle cases of nerve thickening. One other study evaluated qualitative assessment of brachial plexus MRI in a small cohort of patients with CIDP (n = 13) and MMN (n = 10) using a 4-point scale for nerve thickening. 30 They also found a poor interrater reliability evaluated with the interclass correlation coefficient (ICC = 0.47), which corroborates our findings. A qualitative assessment as currently used in clinical practice is, based on these results, not very helpful. Another approach is needed if we want to improve reproducibility and reliability of assessment. Quantification of nerve thickness represents an obvious strategy as earlier nerve ultrasound studies showed excellent test characteristics for the detection of CIDP and MMN. 31,32 In chapter 4 we therefore developed a quantitative measurement method for MRI and calculated its diagnostic value. We measured nerve root sizes at standardized measurement sites (i.e. next to the ganglion and 1 cm distal from this point) in nerve root C5-C7. We found an acceptable reliability (intrarater ICC 0.55 – 0.87; interrater ICC 0.65 – 0.90) and diagnostic value (area under the curve 0.78 – 0.81). Three other studies explored the feasibility of a quantitative assessment of MRI in patients with CIDP (but