As we move towards an era of neuroprotective and disease modifying therapies for neurodegenerative conditions, there is an increasing recognition of the need to better categorize and diagnose patients in prodromal or pre-clinical phases of the disease when such interventions are most likely be effective. Whilst much attention has been focused on Alzheimer’s disease (AD), the last decade has seen significant research investment into the prodromal markers of Lewy body diseases, including Parkinson’s disease and Dementia with Lewy bodies (DLB). Although DLB is the second most common neurodegenerative pathology underlying dementia in the elderly, consensus criteria for the prodromal phase of this disease were only very recently proposed (McKeith et al., Reference McKeith2020).

The criteria for prodromal DLB, which are currently intended for research purposes only, define a Mild Cognitive Impairment (MCI) phase (MCI with Lewy bodies; MCI-LB) as the prototypical prodromal state of DLB. In addition, both a delirium-onset and late psychiatric-onset prodrome of DLB have been proposed but with much less supportive data. Whereas prodromal AD has benefitted from a range of validated biological biomarkers reflecting the underlying neuropathology (i.e. CSF and PET biomarkers of amyloid and tau), widely accepted biomarkers for Lewy body pathology are currently lacking. As such, the identification of prodromal DLB still depends largely on surrogate clinical markers of disease. Accordingly, the diagnosis of MCI-LB requires the presence of MCI—defined according to the National Institute on Aging and Alzheimer’s Association criteria (Albert et al., Reference Albert2011)—alongside a minimum of two “core” clinical features including visual hallucinations, cognitive fluctuations, parkinsonism, and REM sleep behaviour disorder (RBD); or alternatively, one “core” clinical feature and an “indicative” biomarker including reduced dopamine transporter uptake in basal ganglia demonstrated by SPECT or PET, polysomnographic confirmation of REM sleep without atonia (RSWA), and reduced meta-iodobenzylguanidine (MIBG) uptake on myocardial scintigraphy.

Whilst the above criteria stand to advance the field and encourage disease modifying trials in DLB, the identification of patients with prodromal DLB in the clinical setting is still met with several challenges. In practice, DLB is commonly underdiagnosed or misdiagnosed as AD (Nelson et al., Reference Nelson2010). This is accounted for in part by variations in clinical expertise (Vann Jones and O'Brien, Reference Vann Jones and O'Brien2014), but may also reflect the finding that clinically relevant AD pathology co-occurs in up to three-quarters of patients with DLB and can modify the expression of clinical DLB features (Beach et al., Reference Beach2015; Graff-Radford et al., Reference Graff-Radford2017). Meanwhile, DLB pathology can also occur in approximately half of those patients with AD and is associated with more rapid cognitive decline. Thus, identification of co-existent DLB pathology may also have relevance for AD patients in both clinical practice and trial settings. Unfortunately, as mentioned above, the solution to this is stifled by a lack of standardized in vivo markers of Lewy body pathology. Moreover, the adoption and availability of surrogate objective biomarkers such as dopamine transporter imaging or myocardial scintigraphy are generally limited outside of research studies, and especially so in resource-limited settings. Taken together, there is a clear need for studying practical, clinically relevant biomarkers that can help separate DLB and AD, especially during the prodromal phase where these disorders are potentially less well differentiated.

The article by Thomas et al. (Reference Thomas2022) featured in this issue, aims to address this problem by studying olfactory testing as a tool for discriminating MCI-LB and MCI-AD. Olfactory dysfunction is one of the earliest recognized prodromal features of neurodegenerative disease and has been studied in several large prospective community-based studies. Although, olfactory dysfunction is seen across a range of neurological disorders, including AD, it has been most frequently and consistently associated with Lewy body disorders including Parkinson’s disease (PD), in which it is present in up to 96% of patients (Attems et al., Reference Attems, Walker and Jellinger2014). In the Honolulu-Asia Aging study (Ross et al., Reference Ross2008), olfactory dysfunction was found to precede the onset of PD by at least 4 years. In an enriched cohort of patients with isolated REM sleep behaviour disorder, olfactory dysfunction can also predict an earlier conversion to PD or DLB (Postuma et al., Reference Postuma, Gagnon, Vendette, Desjardins and Montplaisir2011). Hyposmia is even included in the recent probabilistic research criteria for prodromal PD proposed by the International Parkinson and Movement Disorder Society (MDS), in which olfactory dysfunction confers an approximate relative risk of 5 and a positive likelihood ratio of 4 for developing PD (Berg et al., Reference Berg2015). It should be noted that these criteria do not exclude DLB, in line with the recent MDS criteria for PD allowing dementia to accompany the diagnosis of PD and dispensing the previous “1-year rule” (Postuma et al., Reference Postuma2015)—a proposition that has not been universally accepted (Boeve et al., Reference Boeve2016).

Neuropathologically, olfactory dysfunction is related to the deposition of pathological proteins that relate to both AD (tau and amyloid beta) and Lewy body pathology (Attems et al., Reference Attems, Walker and Jellinger2014). Despite occurring in both disorders, olfactory dysfunction has been shown to be useful in a number of studies for differentiating between AD and DLB, with a recent large clinicopathological study showing Lewy body pathology to be associated with severe hyposmia (Beach et al., Reference Beach2020). Given these findings, one would expect hyposmia to be a reasonable discriminating marker between DLB and AD, especially at the earlier MCI stage, perhaps when there has been less opportunity for comorbid pathology to develop. Therefore, translating from the findings in DLB, one would also expect hyposmia to be more severe in MCI-LB, than MCI-AD.

To test these hypotheses, Thomas and colleagues performed a cross-sectional study of a community-dwelling population of patients recruited from local memory clinics and diagnosed them according to the most recent consensus research criteria with MCI-LB (n = 38 probable MCI-LB and n = 19 possible MCI-LB), MCI-AD (n = 33), and age-matched controls (n = 32). Olfaction was assessed using the Sniffin’ Sticks test, which is one of the two most commonly used methods to test olfaction in studies of Lewy body disorders. The Sniffin’ Sticks test (Burghardt, Wedel, Germany), is an investigator-administered test involving the presentation of felt pens impregnated with an odorant and can be used to test olfactory threshold, odour discrimination, and odour identification (Hummel et al., Reference Hummel, Sekinger, Wolf, Pauli and Kobal1997). The kit can be re-used for several months on hundreds of patients. The other method often used to test olfaction is the University of Pennsylvania Smell Identification test, which comes as individual “scratch-and-sniff” odour identification tests (Doty et al., Reference Doty, Shaman and Dann1984). Both tests are simple and easy to administer, with the latter even permitting self-administration.

Using the 16-item Sniffin’ Sticks (the component of the Sniffin’ sticks which mainly assesses odour identification), Thomas and colleagues confirmed that at a group level, olfactory performance was indeed worse in those patients with probable MCI-LB than patients with MCI-AD. Uniquely, they also showed that olfaction scores in a subset of patients with possible MCI-LB (patients with MCI that only possessed one core clinical feature or indicative biomarker of DLB) were better than probable MCI-LB and comparable to patients with MCI-AD, which is consistent with the understanding that they may represent a less differentiated form of MCI, and therefore likely include a proportion of patients with underlying AD pathology.

Olfaction is known to worsen with age, a finding that was replicated in the present study wherein the mean age of the participant groups was between 73.1 and 74.3 years of age. In the largest multicentre study assessing the utility of the SS-16 in PD, an optimal cut-off of 10 or less was found to differentiate PD from controls and atypical Parkinsonism in a specialty clinic setting with high sensitivity and specificity (Mahlknecht et al., Reference Mahlknecht2016). However, in keeping with the known demographic differences between PD and dementia, the mean ages across the groups were 5–10 years younger than those of the present study. Concordantly, Thomas and colleagues found that the previous cut-off had reasonable sensitivity but poor specificity for discriminating MCI-LB from MCI-AD and healthy controls. Consistent with their hypothesis, a lower cut-off of 7 or less was found to distinguish probable MCI-LB with a 63% sensitivity but with improved specificities against MCI-AD (73%) and healthy controls (97%). Thus, in the appropriate clinical setting, significant olfactory impairment defined as an SS-16 score of 7 or below can be helpful as part of a holistic clinical assessment in distinguishing MCI-LB from MCI-AD.

There were two additional interesting exploratory analyses performed in this study. The first demonstrated that the total score on the SS-16 was correlated with cognitive testing scores, though not with motor symptom scores based on the motor subsection of the MDS-Unified Parkinson disease rating scale (MDS-UPDRS). The link with cognition mirrors similar studies in DLB and AD (Yoo et al., Reference Yoo2018). Meanwhile, the lack of correlation with motor symptoms, as noted by the authors in the discussion could reflect milder motor symptoms given the selection of patients according to the “one year rule” – i.e. MCI would have had to precede or occur within one year of motor symptoms. However, the MDS-UPDRS scores in the probable MCI-LB group were reasonably high (median 21 and up to 50). An alternative explanation of this finding may be that it could reflect a true neuropathological relationship. Indeed, reduced striatal dopamine transporter uptake is not universally found in DLB, and can be normal in a quarter of cases (McKeith et al., Reference McKeith2007). A potential differential pathological involvement between striatal (resulting in motor disturbances) and olfactory structures strikes at the heart of ongoing scientific debates regarding our understanding of Lewy body disease progression (see for instance (Borghammer and Van Den Berge, Reference Borghammer and Van Den Berge2019; Brundin and Melki, Reference Brundin and Melki2017; Hawkes et al., Reference Hawkes, Del Tredici and Braak2007). Such discourse stems from Braak’s initial description that pathological involvement of olfactory and lower brainstem structures occurs at the earliest stages of Lewy body disease (Braak et al., Reference Braak, Del Tredici, Rub, de Vos, Jansen Steur and Braak2003), leading to hypotheses linking different subtypes of PD to the initial site of pathology and the pattern of prion-like spread of pathological α-synuclein. Recent work has even supported the proposition of a “brain-first” (pathology originating in the CNS including limbic and olfactory structures) versus “body-first” (pathology originating in the peripheral nervous system) (Borghammer et al., Reference Borghammer2021). Whether similar subtypes exist for DLB and could explain the differential striatal and olfactory involvement is a compelling area of future research.

The second interesting exploratory analysis in the current study relevant to practising clinicians found self-reporting of olfactory dysfunction correlated poorly with objective testing. This was especially so for patients who reported no olfactory dysfunction initially—the majority of whom scored below the cut-off on SS-16 testing. Whilst this may be unsurprising, it should serve as a reminder to clinicians who often rely exclusively on interview reporting to elicit this sign. Absence of self-reported olfactory disturbance perhaps offers little reassurance, whilst the presence of self-recognized olfactory loss may be of more value.

The study by Thomas et al. benefits from a reasonably sized and well-phenotyped cohort of patients, employing several objective biomarkers, and being one of few that have defined their MCI population according to the most recent consensus criteria for prodromal DLB. Differentiation between MCI-LB, MCI-AD, and dementia stages can be challenging and subject to variability. This was overcome with the use of a rigorous expert panel system. Further detail regarding the phenotype of their population including the prevalence of the different core and supportive features (e.g. hallucinations, REM sleep behaviour disorder) and disease duration may have been useful to assess the generalizability across MCI populations. Also, as patients were recruited from memory clinics, there is an unavoidable degree of ascertainment bias, and it is unclear whether the cut-off scores in the study apply across all medical settings including non-specialist practice.

As a cross-sectional study, it is difficult to know with greater certainty if and how the sensitivity and specificity of the proposed cut-offs are dependent on the underlying disease severity and time to conversion to dementia as suggested by the significant correlation between cognitive impairment and the degree of hyposmia. Longitudinal studies will be needed to help clarify this issue in addition to confirming the proportion of patients who convert to the expected diagnosis.

Unravelling the contributions of DLB and AD pathology to olfactory dysfunction in vivo remains an open problem. Previous work published in International Psychogeriatrics has explored the relevance of olfactory deficits in relation to amnestic MCI and AD (Bahar-Fuchs et al., Reference Bahar-Fuchs, Moss, Rowe and Savage2011; Velayudhan et al., Reference Velayudhan, Pritchard, Powell, Proitsi and Lovestone2013; Velayudhan et al., Reference Velayudhan, Wilson-Morkeh, Penney, Jesu, Baillon and Brugha2019). These studies have suggested that olfaction may be a marker of severity and progression in AD (Velayudhan et al., Reference Velayudhan, Pritchard, Powell, Proitsi and Lovestone2013), and as a marker of conversion from MCI to AD (Tahmasebi et al., Reference Tahmasebi, Zehetmayer, Pusswald, Kovacs, Stögmann and Lehrner2019; Velayudhan et al., Reference Velayudhan, Wilson-Morkeh, Penney, Jesu, Baillon and Brugha2019). The current study by Thomas et al., builds on and extends this previous work showing that olfactory dysfunction is more likely to be severe in those cases clinically suspected of having underlying Lewy body pathology compared to AD at the MCI stage. In this regard, it would be interesting to longitudinally follow the subset of MCI-AD patients in the present study with severe olfactory disturbance to see what proportion of these end up converting to MCI-LB or DLB. Recent neuropathological findings would suggest that a proportion will likely have some degree of comorbid Lewy body pathology (Beach et al., Reference Beach2020). On the other hand, it has also been shown that AD pathology can modify the expression of DLB features (Ferreira et al., Reference Ferreira2020), and so it would be suspected that a proportion of patients who start off with MCI-LB may eventually go on to be diagnosed with AD, perhaps explaining why olfactory testing may be less discriminatory and specific once dementia is established. This also makes an argument for clinicians to apply biomarkers such as olfaction to phenotype patients as early as possible in their diagnostic journey, rather than wait until more florid dementia symptoms become apparent.

In summary, the study by Thomas and colleagues suggests that olfactory testing may indeed be a useful adjunct in the phenotyping of patients with MCI and sets the stage for future prospective longitudinal and clinicopathological studies. Whilst not as specific as other more expensive, and potentially less accessible objective biomarkers such as dopamine transporter imaging or MIBG, an objective measurement of impaired olfaction remains a useful signpost and initial step that should alert the clinician towards the potential diagnosis of MCI-LB, and prompt the careful surveillance for other core and supportive features.