Comparing the accuracy and neuroanatomical correlates of the UPSIT-40 and the Sniffin' Sticks test in REM sleep behavior disorder

BACKGROUND

Olfactory impairment increases the risk of developing neurodegenerative diseases in patients with idiopathic REM sleep behavior disorder (IRBD). Knowing the test properties of distinct olfactory measures could contribute to their selection for clinical or research purposes.

OBJECTIVE

To compare the accuracy in distinguishing IRBD patients from controls with the University of Pennsylvania Smell Identification Test (UPSIT-40) and Sniffin' Sticks Extended test, and to assess the gray-matter volume correlates of these tests.

METHOD

Twenty-one patients with IRBD and 27 healthy controls were assessed using both olfactory tests. Independent logistic regressions were computed with diagnosis as a dependent variable and olfactory measures as predictive variables. Receiver operating characteristic curves were computed for each olfactory subtest. Diagnostic accuracy for IRBD was calculated according to the resulting optimal cut-off score. Structural MRI data, acquired with a 3T scanner, were analyzed with voxel-based morphometry.

RESULTS

Patients differed from controls in all olfactory measures. The Sniffin-Identification correctly classified 89.1% of cases; the UPSIT-40, 85.4%; the Sniffin-Discrimination, 82.6%; the Sniffin-Total, 81.8%; and the Sniffin-Threshold, 77.3%. Respective AUROC, optimal cut-off, sensitivity, and specificity for each test were: 0.902, ≤26, 85.7%, and 85.2% for the UPSIT-40; 0.884, ≤29, 89.5%, and 76.0% for the Sniffin-Total; 0.922, ≤11, 90.5%, and 88.0% for the Sniffin-Identification; 0.739, ≤4, 73.7%, and 76.0% for the Sniffin-Threshold; and 0.838, ≤11, 85.7%, and 76.0% for the Sniffin-Discrimination. UPSIT-40 scores correlated with gray-matter volumes in orbitofrontal regions in anosmic patients.

CONCLUSIONS

UPSIT-40 and Sniffin' Identification showed similar discrimination accuracy, but only the UPSIT-40 showed structural correlates (p ≤ .05 FDR-corrected).