Individual-reader diagnostic performance and between-reader agreement in assessment of subjects with Parkinsonian syndrome or dementia using 123I-ioflupane injection (DaTscan) imaging.

Abstract:

UNLABELLED: Establishing an early, accurate diagnosis is fundamental for appropriate clinical management of patients with movement disorders or dementia. Ioflupane (123)I Injection (DaTscan, (123)I-ioflupane) is an important adjunct to support the clinical diagnosis. Understanding individual-reader diagnostic performance of (123)I-ioflupane in a variety of clinical scenarios is essential. METHODS: Sensitivity, specificity, interreader, and intrareader data from 5 multicenter clinical studies were reviewed. The different study designs offered an assortment of variables to assess the effects on the diagnostic performance of (123)I-ioflupane: on-site versus 3-5 blinded image readers, number of image evaluations, early/uncertain versus late/confirmed clinical diagnosis as reference standard, and subjects with movement disorders versus dementia. RESULTS: Eight hundred eighteen subjects had individual-reader efficacy data available for analysis. In general, sensitivity and specificity were high and comparable between on-site versus blinded independent readers. In subjects with dementia, when the clinical diagnosis was made at month 12 versus baseline, specificity improved from 77.4%-91.2% to 81.6%-95.0%. In subjects with movement disorders, this effect was observed to an even greater extent, when diagnostic performance using month-18 diagnosis as a reference standard (sensitivity, 67.0%-73.7%; specificity, 75.0%-83.3%) was compared versus month-36 diagnosis (77.5%-80.3% and 90.3%-96.8%, respectively). Diagnostic performance was similar in subjects with dementia (74.4%-89.9% and 77.4%-95.0%, respectively) and subjects with movement disorders (67.0%-97.9% and 71.4%-98.4%, respectively). In most of the comparisons, between-reader agreement was very good (almost perfect), with κ ranging from 0.81 to 1.00. Within-reader agreement, measured in 1 study, was 100% for 3 blinded readers. CONCLUSION: Individual-reader diagnostic performance, as assessed by measuring sensitivity and specificity of (123)I-ioflupane to detect the presence or absence of striatal dopaminergic deficit, using the clinical diagnosis as a reference standard, was high in subjects with either movement disorders or dementia and was similar in on-site readers versus blinded analyses. Between- and within-reader agreements were very good (almost perfect). Longer follow-up between imaging and clinical diagnosis improved the diagnostic accuracy, most likely due to improvement in the clinical diagnosis reference standard, rather than changes in reader accuracy.