Abstract
Smell dysfunction is a common and underdiagnosed medical condition that can have serious consequences. It is also an early biomarker of Alzheimer’s disease that precedes detectable memory loss. Clinical tests that evaluate the sense of smell face two major challenges. First, human sensitivity to individual odorants varies significantly, leading to potential misdiagnosis of people with an otherwise normal sense of smell but insensitivity to the test odorant. Second, prior familiarity with odor stimuli can bias smell test performance. We have developed new non- semantic tests for olfactory sensitivity (SMELL-S) and olfactory resolution (SMELL-R) that overcome these challenges by using mixtures of odorants that have unfamiliar smells. The tests can be self-administered with minimal training and showed high test-retest reliability. Because SMELL-S uses odor mixtures rather than a single molecule, odor-specific insensitivity is averaged out. Indeed, SMELL-S accurately distinguished people with normal and dysfunctional smell. SMELL-R is a discrimination test in which the difference between two stimulus mixtures can be altered stepwise. This is an advance over current discrimination tests, which ask subjects to discriminate monomolecular odorants whose difference cannot be objectively calculated. SMELL-R showed significantly less bias in scores between North American and Taiwanese subjects than conventional semantically-based smell tests that need to be adapted and translated to different populations. We predict that SMELL-S and SMELL-R will be broadly effective in diagnosing smell dysfunction, including that associated with the earliest signs of memory loss in Alzheimer’s disease.
Significance statement Currently available smell testing methods can misdiagnose subjects with lack of prior experience or insensitivity to the odorants used in the test. This introduces a source of bias into clinical tests aimed at detecting patients with olfactory dysfunction. We have developed smell tests that use mixtures of 30 molecules that average out the variability in sensitivity to individual molecules. Because these mixtures have unfamiliar odors, and the tests are non-semantic, their use eliminates differences in test performance due to the familiarity with the smells or the words used to describe them. The SMELL-S and SMELL-R tests facilitate smell testing of diverse populations, without the need to adapt the test stimuli.
