More than a thousand elderly people without dementia were recruited from the community in a specific area of North Manhattan.  Participants were Medicare recipients, with a mean age of 80 years. In the longitudinal study described by Devangere Devanand, Division of Geriatric Psychiatry, Columbia University, New York, USA, olfaction was assessed at baseline using the University of Pennsylvania Smell Identification Test (UPSIT).

The UPSIT consists of a series of “scratch and sniff” strips which release well-known smells at suprathreshold levels when rubbed with a pencil. For each strip, subjects select the smell they identify  from a set of four alternatives.  Performance on the test – which is scored out of 40 – are compared against norms for age and gender.

Participants also completed the Selective Reminding Test of verbal learning and memory. Follow-up data at two or four years are available on 757 participants.

After adjustment for covariates, lower scores on UPSIT were significantly associated with cognitive decline over that period. Scores on the verbal learning test were not.

During follow-up, 109 participants in the study developed dementia: in the case of 101 people, the diagnosis was AD.  Development of AD was significantly associated with lower olfaction scores at baseline, even after adjustment for relevant demographic, cognitive and functional covariates.

It may be fairer to say that good olfaction makes it likely that a person will not develop dementia, rather than that poor olfaction makes it likely that they will

Poor smell also predicts overall mortality

Over a median four years’ follow up of the same sample of North Manhattan residents, 30% died. As well as correlating with risk of dementia, poorer olfaction at baseline also was also highly correlated with risk of death. This was so even when relevant covariates -- including dementia itself, and other comorbidities and risk factors – were taken into account.

The mortality risk in the lowest quartile of UPSIT scores was 3.8 times that among people in the quartile with the best olfaction scores. Why there should be such a strong relationship with death rate is not yet clear.

Neurofibrillary tangles in the olfactory bulb

We have known for some years that the olfactory bulb is affected early in AD, when numerous neurofibrillary tangles are evident. People with established AD consistently show deficits on a “scratch and sniff” task requiring them to identify and name strong and frequently encountered smells such as cheese, chocolate and apple.

Performance on the task may be compromised by a history of smoking or a current upper respiratory tract infection. But it seems that deficits are more in the naming than in the identification, ie not so much in the nose as in the brain. However, similar deficits are also found in Parkinson’s disease and schizophrenia, so the test is not very specific.

Deficits in smell identification more in the brain than the nose

In a study of people with mild cognitive impairment, those with a low olfaction score were four times more likely to develop AD during follow-up over three years than those who were not impaired.  The greater the olfactory abnormality, the greater the AD pathology in the brain.  Even so, poor olfaction was only one of many relevant biomarkers, just as lipid level might be in cardiovascular disease.

Searching for dementia biomarkers

Potential biomarkers range from neuropsychological tests, through functional impairments identified by caregivers, to amyloid imaging on PET. Olfactory deficits can now be added to the list.

The full 40-item UPSIT battery can be completed in twenty minutes, and a slimmed down 12-item version in six or seven. So the test is practical.

Even so, it may be fairer to say that a good olfaction score makes it likely that a person will not develop dementia, rather than that a high score makes it likely that they will, Dr Devanand told the Atlanta meeting. In the longitudinal study of elderly people without dementia, the risk of developing dementia in those who scored 35 out of 40 in the smell test was only 2-4%.