Article Text

Download PDFPDF

Back on the scent: the olfactory system in CNS demyelinating diseases
  1. Albert Joseph1,
  2. Gabriele C DeLuca2
  1. 1Foundation School Offices, St Mary's Hospital, London, UK
  2. 2Nuffield Department of Clinical Neurosciences, John Radcliffe Hospital, Oxford, UK
  1. Correspondence to Professor Gabriele C DeLuca, Nuffield Department of Clinical Neurosciences, Level 1 West Wing, John Radcliffe Hospital, Oxford OX3 9DU, UK; gabriele.deluca{at}


Olfactory dysfunction is recognised across an ever broadening spectrum of neuropsychiatric conditions including central nervous system (CNS) demyelinating diseases such as multiple sclerosis (MS) and neuromyelitis optica (NMO). In this review, we unravel the striking evidence highlighting how olfactory loss is a common clinical feature in MS and NMO. We provide an overview of the supportive psychophysical, electrophysiological, radiological and pathological data that point to the anatomical substrate of olfactory deficits in these diseases. The pattern of underlying pathology affecting the olfactory system is shown to be complex, involving multiple structures that are affected in different ways throughout the course of the disease. This review is the first to synthesise the expanding body of literature on the topic, provides novel insight into the way in which the olfactory system is affected in CNS demyelinating diseases, and raises intriguing questions about the role of this system in the pathogenesis of these diseases.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Olfaction is the most ancient of the senses. Its evolutionary importance is highlighted by the vastness of the olfactory genome, extending over more than 1000 genes in humans. While olfaction is no longer critical to survival in modern times, smell dysfunction can significantly impair quality of life and increase risk to hazardous events, such as gas leaks and fires.1 The association of smell loss with significant psychiatric comorbidity,2 and increased 5-year mortality in older adults,3 further accentuates olfactory impairment as either a cause or consequence of underlying neuropsychiatric disease. Therefore, it is not surprising that this symptom, often overlooked by patients and clinicians, is receiving increasing attention.

Olfactory dysfunction is a recognised feature of a constellation of neuropsychiatric diseases, namely Parkinson's,4 Alzheimer's,5 motor neurone disease,6 depression,7 schizophrenia,8 bipolar disorder9 and anorexia nervosa.10 This spectrum of olfactory involvement continues to broaden with emerging evidence to suggest that smell loss is a feature of central nervous system (CNS) demyelinating diseases, such as multiple sclerosis (MS).11 This review aims to unravel the basic anatomy of the olfactory system and the anatomic substrate of olfactory dysfunction, the tools used in the clinic to assess its impairment, and typical manifestations and presentations of smell loss in MS and other CNS demyelinating disorders. The relevance of olfactory disturbance on the pathogenesis of CNS demyelinating disease will be highlighted.

Olfactory anatomy

Olfactory anatomy is complex with peripheral and CNS contributions critical to function. To fully understand how the olfactory system is affected in neurological disease, an overview of the olfactory processing pathway is necessary. The three major components of olfactory anatomy, peripheral olfactory apparatus, olfactory bulb/tract and olfactory cortex will be summarised here.

  1. Peripheral olfactory apparatus: volatile odorants are inhaled through the nostrils and pass through a cleft in the medial part of the superior turbinate to enter the nasal attic where they are absorbed into a layer of mucous secreted by abundant Bowman's glands. Odorants subsequently encounter the olfactory neuroepithelium, which contains the bipolar cell, the first order sensory neuron of the olfactory system where signal transduction of odorant input into electrical signals takes place. The axons of the neuroepithelium coalesce into ∼50 nerves, which subsequently pass through clefts in the cribriform plate to synapse onto the olfactory bulb of CNS.

  2. Olfactory bulb/tract: the olfactory bulb has a volume of roughly 140 mm3 in the healthy adult and is bordered by the skull base inferiorly and the inferofrontal cortex superiorly. The structure is the site of significant olfactory signal processing, being replete with GABA and dopaminergic interneurons, and contains the anterior olfactory nucleus, a part of the olfactory brain. The bulb contains a unique population of cells, the olfactory ensheathing cells, which play a role in neurogenesis and remyelination that extends into adulthood.12 It is in the olfactory bulb where the second order olfactory neurons extend into the olfactory tract to transmit their axons to the olfactory cortex where further processing takes place.

  3. Olfactory cortex: it is important to make the distinction between the olfactory brain and the olfactory cortex. The former includes all brain regions in which olfactory processing occurs, including both olfactory cortex and grey matter nuclei in the olfactory bulb. Olfactory cortex specifically refers to a constellation of processing modules distributed predominantly throughout the mediotemporal and inferofrontal cortices, with right inferofrontal cortex being typically more involved in olfactory processing than the left.13 More distant cortical structures, such as orbitofrontal cortex and cerebellum, are also involved in olfactory function.14

Olfactory structure is intrinsically linked to its function. However, the precise role these different component parts play in smell loss seen in neurological disease is difficult to ascertain. This is, in part, related to functional interdependence of these regions in olfactory processing and the insensitivity of diagnostic tests to tease out the relative contribution of these anatomic structures to olfactory loss. That being said, clinical and paraclinical tools used to assess olfactory function have cast light onto the roles of the peripheral neuroepithelium, olfactory bulb/tract and olfactory cortex on the ability to smell in health and in demyelinating disease.

Olfactory function and its assessment

Olfactory dysfunction can be manifested in several ways. Anosmia and hyposmia (also referred to as a microsmia) describe complete and partial loss of smell, respectively. Parasomia refers to the inability to correctly identify specific odorants and phantosmias are olfactory hallucinations. Olfaction comprises three major functional components—olfactory detection threshold, identification and discrimination—which are defined and explained in more detail below.

  1. Olfactory detection threshold: minimum concentration at which a given odorant can be detected.15 This component of smell is frequently affected in disease of the nasal apparatus and olfactory neuroepithelium,15 ,16 but is also altered by generalised neurodegeneration such as that seen in Alzheimer's disease (AD).17

  2. Olfactory identification: the ability to specifically label, either verbally or non-verbally, a given odorant at a suprathreshold level.15 Identification is thought to be mediated by neural pathways linking the olfactory bulb and tract to temporal cortices (the seat of olfactory semantic memory). Reductions in olfactory bulb and tract volume have been found to reduce olfactory identification across a number of studies.18 A temporal lobe contribution to olfactory identification is exemplified by epilepsy syndromes with a temporal lobe focus, where explicit identification of smells is a feature. Frontal lobe function has also been implicated.19

  3. Olfactory discrimination: The ability to correctly distinguish between differing odorants at suprathreshold levels.20 Discriminatory function is believed to be especially dependent on frontal cortical activity, requiring attention and executive function to compare and contrast multiple inputs simultaneously.

Psychophysical tests provide objective measurement of overall olfactory function and of its various components. There is a wide range of commercially available tests, the most common being Sniffin’ Sticks20 and University of Pennsylvania Smell Identification Test (UPSIT).21 Sniffin’ Sticks measure Threshold (T), Discrimination (D) and Identification (I) in order to create a composite ‘TDI’ score for overall olfactory function.20 Threshold is measured in a ‘single staircase protocol’ in which the testee is presented with three sticks, two of which contains a solvent and the other a solvent plus an odorant, and has to pick out which contains the odorant. The concentration of the odorant is increased as the test progresses and the number of times that the testee accurately detects the correct stick is used to calculate a Threshold score. An Identification score is generated by presenting the testee with sticks containing specific odorants and calculating frequency at which odorants can be correctly assigned to a given verbal or non-verbal label. The Discrimination score is calculated by the number of times a testee correctly selects the odd one out from a selection of three odorant containing sticks, two of which contain the same odorant. The UPSIT is a pure test of identification.21 The testee is provided a booklet of multiple choice questions ‘for example, this odour smells most like: (1) chocolate; (2) banana; (3) onion or (4) fruit punch’ with a corresponding strip of microencapsulated crystals containing the odorant of interest. The benefits of psychophysical tests are that they are economic, quick and easy to administer. Their drawback is their limited capacity for discerning the anatomical/pathological basis of a given presentation of smell loss.

Electrophysiological tests provide an alternative to olfactory psychophysics. To record an olfactory evoked potential (OEP), the participant is exposed to either gaseous H2S (an olfactory stimulator) or CO2 (a trigeminal stimulant) in a manner that does not alter the thermal or chemical environment of the nose.22 Electrodes at Cz and Pz are simultaneously used to record cerebral responses.22 Increase in latency between stimulation and cerebral response are thought to reflect myelin loss within olfactory pathways, including bulb and tract. In contrast, reductions in amplitude reflect axonal loss in these structures.

Psychophysical and electrophysiological techniques have enabled measurement of the frequency and nature of olfactory disturbance in CNS demyelinating diseases. Data derived from these techniques combined with those obtained from radiological and pathological studies allow the evaluation of the specific anatomic substrate(s) for loss of smell in these diseases, as outlined below.

Olfactory dysfunction in CNS demyelinating diseases

Olfactory disturbance in MS and other CNS diseases where demyelination is a feature, such as neuromyelitis optica (NMO) and acute disseminated encephalomyelitis (ADEM), has been relatively neglected in clinical and research studies. While this arises from this clinical symptom being often overlooked, a key factor lies in a few key historical reports, which suggested that the olfactory system is spared in MS.23–26 It is only in the past two decades that the nature and extent of olfactory disturbance in MS and other demyelinating diseases are being appreciated. The next section of the review will dissect the clinical, electrophysiological, radiographic and pathological evidence for olfactory involvement in MS, NMO and ADEM, where available.

In order to systematically review literature relating to olfactory dysfunction in the context of these conditions, searches for all combinations of the following words were performed on both Google Scholar and PubMed up to 11 October 2015: ‘multiple sclerosis’, ‘MS’, ‘neuromyelitis optica’, ‘NMO’, ‘Devic's’, ‘Acute Disseminated Encephaolmyelitis’, ‘ADEM’, ‘Demyelination’, ‘Demyelinating’ with either ‘smell’, ‘chemosensory’, ‘olfaction’, ‘olfactory’, ‘odor’, ‘odorant’, ‘sinusitis’, ‘rhinosinusitis’, ‘sinus’. Relevant literature was read and included in this review if deemed relevant. If the article contained data relating to frequency of smell loss in demyelinating disease, then it was included in the online supplementary table S1.

Supplementary table

Summary of Studies of Olfactory Dysfunction in MS and NMO

Clinical data

The weight of available evidence suggests that olfactory symptoms are common in MS although reports have been conflicting. First noted in 1971, Wender and Szmeja27 observed that olfactory loss occurred in 35% of patients with MS compared with controls. Despite this remarkable finding, attention was drawn away from further pursuit of olfactory disturbance in MS after a key paper published by Ansari23 which refuted smell loss in the disease. Around that time, an experimental report on human postmortem tissue identified a paucity of myelin basic protein in the olfactory bulb/tract,24 suggesting decreased likelihood of these CNS structures being subject to immune attack in MS, adding further support to this clinical finding. While a report in 1977 by Pinching28 once again identified olfactory disturbance in patients with MS and challenged the claim that there was a paucity of myelin within the bulb/tract, it was only from the late 1980s that momentum gathered in the investigation of loss of smell in MS. To date, a total of 30 studies on olfactory function in MS and other demyelinating disease have been published.19 ,22–24 ,27–53 The message that smell loss not only occurs in MS, but is common is now clear (see online supplementary table S1).

Psychophysical testing has found the mean frequency of olfactory disturbance in patients with MS to be 35.6% (11–50%), with considerable variability in the nature and extent of disturbances noted between studies. In part, this reflects studies that have assessed patients across the temporal and severity continuum of the disease. In MS, the most striking difference in psychophysical findings exists between early versus late disease stages. Olfactory loss can be detected at the earliest of disease stages, namely at first presentation and in early relapsing-remitting (RR) MS.41 ,44 ,47 Lutterotti et al41 observed olfactory impairments in 4/5 patients with clinically isolated syndrome. Similar findings were found in a cohort presenting with monosymptomatic episodes of acute optic neuritis.47 This patient group was also reported to have heightened awareness of their olfactory troubles relative to other disease stages.47 ,51 ,52 In clinically isolated syndrome and early RRMS, increased disease activity within the preceding 2 years of olfactory assessment was associated with poorer olfactory function, especially in the threshold domain.41 ,44 In a similar vein, relapses negatively impact olfactory function, again with threshold disturbance being preferentially affected.41 With disease evolution, olfactory impairment appears to increasingly affect other olfactory domains, namely identification and/or discrimination.19 ,35 ,41 ,44 ,49

Olfactory disturbance is also well documented in progressive MS. Approximately 60–90% of patients with progressive MS have objective olfactory loss.47 ,50 Not only is olfactory loss more common, it has been shown to be more severe in progressive and more advanced disease.19 ,35 ,37 ,41 ,44 ,47 ,49 A common theme across studies in patients with MS with progressive disease is impairment of olfactory discrimination and identification,19 ,35 ,44 ,49 similar to that seen in late RRMS.41 ,42 ,44 The shift in the nature of olfactory disturbance between early and late disease stages raises the question as to whether different olfactory structures are affected at different disease stages.

Psychophysiological data on olfactory function in other demyelinating diseases is sparse. NMO has been linked to smell impairment. Schmidt et al48 report olfactory loss in 50% (5/10) of their NMO cohort, and Revis et al43 found olfactory identification to be significantly more impaired in NMO than in MS. In a larger study, Zhang et al53 observed smell dysfunction in 53% (26/49) of patients with NMO spectrum disorder. The frequency and extent to which olfactory dysfunction forms part of the clinical spectrum of other CNS demyelinating diseases, such as ADEM, myelin oligodendrocyte glycoprotein encephalomyelitis, and other immune-mediated CNS disorders, is unknown and warrants further investigation.

Electrophysiological data

Electrophysiological data have added an important objective element to olfactory testing. OEPs are deranged in MS.22 ,29 ,31 ,38 ,39 In a cohort of 45 patients with MS and 47 controls, Hawkes et al found an increase in latency and a decrease in amplitude of the H2S (olfactory)-related OEP in MS compared with controls, suggesting demyelination and axonal loss within olfactory bulb/tract and/or olfactory brain as an element of MS pathology.22 ,38 The extent of OEP disturbance in MS related to poor performance on the UPSIT score and to the extent of physical disability, as measured by Expanded Disability Status Score (EDSS), in the disease (see clinical association section below).22 ,38 Other independent groups have corroborated these OEP findings in MS, adding further credence to physiological disturbance of these olfactory structures at some point in the clinical evolution of the disease.29 ,31 ,39

The value of OEPs as a diagnostic tool in MS was also evaluated by Hawkes et al. While visual evoked potentials (VEPs) were more commonly abnormal than OEPs, there was a subgroup of patients within the studied cohort exhibiting deficits in OEP/UPSIT but not in relation to VEP, suggesting that addition of olfactory assessment might increase overall sensitivity of diagnostic investigations.22 ,38

Radiological data

MRI has provided an excellent means of identifying structural correlates of olfactory dysfunction in patients with MS. However, the small size of the olfactory bulbs and tracts and MRI signal artefacts generated by the sinuses in close proximity to these structures have biased early studies to evaluate changes in the olfactory brain. In a landmark study, Doty et al32 ,33 demonstrated a significant positive correlation between reductions in UPSIT score with T2 lesion burden in the inferofrontal and medial temporal cortices of 26 patients with MS, 10 of whom exhibited hyposmia (figure 1A,B). This study was of particular significance because the authors highlighted that it was the first study to link clinical with radiological abnormalities. Importantly, the contribution of olfactory bulb/tract was not considered.32 ,33 Several other clinicoradiographic studies have supported this work,19 ,34 ,49 ,51 ,52 although two reports with large, well-characterised MS cohorts failed to replicate these findings.42 ,44 Differences in MS demographic and clinical features (ie, disease course, duration and severity), olfactory assessment techniques used, and MRI resolution may explain these discrepancies. The application of diffusion-weighted MRI protocols has increased sensitivity to detect subtle pathological changes in the olfactory brain, which have been shown to correlate more significantly with reduced metrics of olfactory function, as measured by TDI.19 ,35 Temporal factors undoubtedly influence the relationship between olfaction and pathology in the MS brain. A longitudinal study over a 18–20-month period found olfactory function fluctuated in a manner which mirrored disease activity in the olfactory brain adding support to this claim.34

Figure 1

MRI studies: the olfactory cortex and olfactory bulb in multiple sclerosis (MS) and relation to smell dysfunction. Smell loss (as measured by University of Pennsylvania Smell Identification Test (UPSIT)) correlates with the extent of measures of MRI MS-related lesion burden in cortical regions responsible for olfaction (ie, inferior and temporal cortices; A) with such a relationship between smell loss and lesion burden in other (non-olfactory) brain regions not seen (B) (from Doty et al32 with permission). This seminal finding pointed to the olfactory brain as the main substrate of smell loss in MS. This view has been recently broadened to include a role for olfactory bulb pathology as an important contributor to smell loss in MS with lower mean olfactory bulb (bulbus olfactorius) volume observed in hyposmic compared with normosmic patients with MS (C; from Holinski et al39 with permission).

Recent MRI studies have examined olfactory bulbs and tracts as possible contributors to olfactory disturbance in MS.37 ,39 ,49 Through the measurement of olfactory bulb and tract volume, a Berlin-based group have repeatedly demonstrated reductions in these structures in MS compared with controls,37 ,39 ,49 and have shown that these changes relate to loss of smell and37 ,49 OEP disturbance (figure 1C).39 Olfactory bulb volume and olfactory sulcus depth have also been shown to be inversely correlated with disease duration and attack frequency in MS.54 Similarly, olfactory loss in NMO has been linked to reductions in bulb volume, alongside grey matter reduction within the piriform gyri.53 The implication of these studies is substantial. Given that atrophy is likely a pathological surrogate of loss of myelin and/or axons, the demonstrated involvement of olfactory bulb/tract has challenged the view that the olfactory brain is solely responsible for olfactory deficits often encountered in these diseases. Pathological studies engender further support to this claim (see below).

Pathological data

As outlined in the previous sections of this review, the conflation of clinical and radiographic data confirms that olfactory dysfunction is a common feature in MS. However, the relative non-specificity of clinical, paraclinical and radiographic tools limit their ability to decipher adequately the substrate of olfactory loss in MS. Pathological studies carefully evaluating postmortem human tissue (olfactory bulb/tract and brain) have filled this important knowledge gap.11 ,55–57 While involvement of the olfactory bulb and tract has been only recently confirmed,11 awareness of olfactory bulb and tract pathology in MS dates back to the 19th century. In his careful monograph entitled ‘Manual of Diseases of the Nervous System’, Gowers55 recognised the frequency of olfactory bulb/tract pathology noting, “…the nerve may be grey in its entire thickness or in part”. Not surprisingly, his brief description was neglected for several decades thereafter. The failure of eminent neuropathologists contemporary to Gowers, such as Cruveilhier,58 Charcot59 and Dawson,60 to recognise olfactory bulb/tract pathology certainly contributed to the lack of interest in these structures in MS pathology. Zimmerman and Netsky's26 failure to detect demyelination in the olfactory bulb/tract in MS postmortem cases, and the (inaccurate)28 claim that these structures have a lower density of myelin basic protein,25 further directed attention away from these anatomic components of the olfactory system.23 In fact, for a number of years, the misconception that these CNS structures are ‘spared’ in MS abounded in scientific circles. A report in German,57 and a fleeting reference to unpublished observations of olfactory bulb/tract pathology in MS,56 certainly did not help dispel this belief.

A rekindled interest in the role of the olfactory bulb and tract in MS pathology has surfaced in recent years. Fundamental observations derived from clinical, electrophysiological and radiographic studies inspired careful pathological re-examination of olfactory bulb and tract involvement in MS and other CNS demyelinating diseases. DeLuca et al11 evaluated these olfactory structures in a large postmortem cohort of MS (n=17), NMO (n=3) and ADEM (n=7) cases and compared them to herpes simplex encephalitis (HSE; n=3), AD (n=4) and non-neurological controls (n=8). Olfactory bulb/tract demyelination was frequent in all demyelinating diseases (MS 12/17 (70.6%); ADEM 3/7 (42.9%); NMO 2/3 (66.7%)) but was absent in HSE, AD and non-neurological controls (figure 2).11 The frequency of olfactory bulb/tract demyelination in MS is similar to other areas of the CNS preferentially affected in the disease, such as the optic nerve, corpus callosum, periventricular white matter and cervical spinal cord. Inflammation was greater in the demyelinating diseases compared with non-neurological controls.11 Olfactory bulb/tract axonal loss was most severe in MS where it correlated significantly with the extent of demyelination (r=0.610, p=0.009) and parenchymal inflammation (r=0.681, p=0.003).11 This study was the first to highlight the distribution of aquaporin (AQP)-4 expression in the human olfactory bulb and tract in non-neurological controls and how its expression is altered in demyelinating diseases. The loss of AQP-4 in NMO lesions within the olfactory bulb/tract was consistent with the pattern seen in NMO lesions elsewhere in the nervous system (figure 3). The extent of remyelination and neurogenesis in the olfactory bulb/tract of these CNS demyelinating diseases (and related animal models) remains unexplored and would be worthy of investigation.

Figure 2

Pathological evidence of demyelination within olfactory bulb/tract in multiple sclerosis (MS). Olfactory bulb and tract demyelination in demyelinating diseases compared with non-neurological controls. (A–H) Olfactory bulbs and tracts stained for myelin (proteolipid protein). High power views of boxed areas represent gray matter (A, C, E and G) and white matter (B, D, F and H) regions for each disease category. In controls, olfactory gray matter structures demonstrated a fine reticulated meshwork of myelin (A). In contrast, myelin in olfactory white matter structures (lateral, medial and intermediate striae) was more densely packed and linearly organised. In each of the demyelinating diseases (MS (C and D), acute disseminated encephalomyelitis (ADEM; E and F) and neuromyelitis optica (NMO; G and H)), olfactory bulb and tract demyelination affected both gray and white matter regions, the extent of which was consistently most severe in MS. Scale bars represent 100 μm. Figure and figure legend adapted from DeLuca et al11 with permission.

Figure 3

Aquaporin-4 (AQP-4) staining in non-neurological controls and demyelinated lesions. AQP-4 staining derived from regions outlined in corresponding myelin-stained (proteolipid protein) sections of the olfactory bulbs and/or tracts (A–F). In non-neurological controls (A–C), aquaporin-4 staining is located throughout the parenchyma in both white (B) and grey (magnified view of small box on olfactory tract in C) matter, being significantly more intense in the former especially in the subpial zone (A and with magnified view of box area in B). Intense staining was also noted around blood vessels. There is selective loss of aquaporin-4 staining within neuromyelitis optica (NMO) lesions (E) compared to lesions in multiple sclerosis (MS; D) and acute disseminated encephalomyelitis (ADEM; F) where AQP-4 staining is increased. Scale bars represent 1 mm in (A), and 50 μm in (B–F). Figure and figure legend adapted from DeLuca et al11 with permission.

Relationship between disease in olfactory bulb and tract and its neighbouring structures

Recent observations suggest that olfactory bulb and tract pathology might relate to that seen in neighbouring olfactory brain structures. Anatomically, the olfactory bulb and tract are virtually juxtaposed to the inferior frontal cortex, with only a fine meningeal layer separating these structures.11 Given what is known about the tight relationship between cortical demyelination and meningeal inflammation,61 it might be inferred that inflammatory demyelination of one structure relate to that seen in the other. DeLuca et al11 explored this possibility by evaluating the relationship between olfactory bulb/tract demyelination and that seen in the inferior frontal cortex in MS and ADEM, diseases where cortical demyelination is an established feature. These investigators showed that the extent of olfactory bulb/tract demyelination correlated with that found in the adjacent inferior frontal cortex.11 To exclude the possibility, these relationships were not merely a reflection of global disease burden, the extent of demyelination in the more distant hippocampus of these MS and ADEM cases was not found to correlate with that found in olfactory bulb and tract (figure 4).11 These pathological findings are also not without radiographic support. An MRI study of 30 patients with MS quantified olfactory bulb/tract atrophy and found that its extent correlated significantly with T2 lesion burden and atrophy in the anatomically proximate olfactory brain but not in non-contiguous brain regions.49 Imaging studies relating olfactory bulb/tract and cortical pathology in ADEM are lacking. While cortical demyelination is not thought to be a feature of NMO,62 recent evidence suggesting the occurrence of cortical neuronal loss in the absence of astrocyte loss and secondary demyelination in this disease support the cortical contribution to olfactory loss observed in NMO by Zhang et al.63

Figure 4

Relationship between the olfactory bulb/tract and olfactory cortex in central nervous system demyelinating disease. Inferofrontal cortical demyelination was significantly greater in cases with high levels of olfactory bulb and tract demyelination (red bar) compared with cases with low levels (blue bar) in demyelinating diseases where cortical demyelination is a feature (ie, multiple sclerosis and acute disseminated encephalomyelitis). No such relationship was found between the extent of olfactory bulb/tract and hippocampal demyelination (From DeLuca et al11 with permission; A). Similarly, in (B), olfactory bulb volume (here called bulbus olfactorius (OB)) correlated with the volume of lesions in the olfactory cortex (olfactory brain (BO); from Holinski et al39 with permission). These findings suggest that olfactory bulb/tract and adjacent cortical demyelination are associated with one another highlighting their combined importance in smell loss in demyelinating disease.

Linking olfactory anatomy to substrate of loss of smell in demyelinating disease

Pathological and radiographic studies have provided a detailed topographical map of pathology along the olfactory system in CNS demyelinating diseases. However, there are limitations in separating out the contributions of the components of olfactory anatomy to loss of function in olfactory threshold, discrimination and identification. The reasons for this are multiple, including a relative lack of sensitivity and specificity of the batteries used to assess olfaction because of the interdependence of olfactory domains (ie, threshold, discrimination and identification) on one another, and the inherent subjective nature of most investigative tools used to assess loss of smell. That being said, studies of olfactory dysfunction in demyelinating disease can provide insight into which olfactory structures are affected at various time points (ie, early vs late), clinical stages (ie, RR vs progressive), and how smell loss relates to other symptoms often encountered in the disease (ie, neuropsychiatric and disability measures). As outlined above, loss of smell seems to differ in early compared with late disease stages. Clinically isolated syndromes, which represent the first clinical manifestation of demyelinating disease, have a surprisingly high frequency of smell loss, with patients typically being more subjectively aware of the olfactory deficits.47 In particular, olfactory threshold seems to be prominently affected at this early disease stage, with relative sparing of olfactory discrimination and identification.41 ,44 The anatomic correlate of this observation is not clear. However, the link between olfactory threshold and neuroepithelial and cortical function implicates early involvement of one or both of these structures.14–17 The relative sparing of discrimination and identification function, olfactory domains dependent on olfactory cortex, challenges a contribution of cortical pathology to early threshold impairment in MS.41 ,44 The possibility that olfactory neuroepithelium could be affected in the early stages of disease is intriguing given epidemiological and radiographic evidence linking nasal sinusitis with MS and AON.64–67 That obstructed aeration induced by nasal sinusitis could explain altered olfactory threshold that occurs early on is made less likely by the fact that olfactory studies in early MS have excluded patients with active or chronic sinus disease. The implication is that the olfactory neuroepithelium is pathologically affected in early MS, which is intriguing from a pathogenesis point of view (see below). This does not exclude the possibility that olfactory submodalities (ie, threshold, discrimination and identification) subserved by olfactory cortex may have different levels of susceptibility to injury. In NMO, the association of both anti-AQP4 seropositivity and radiological abnormalities within the olfactory bulb and cortex with olfactory dysfunction is explained by the diffuse expression of AQP-4 throughout the olfactory system.53 Involvement of the olfactory neuroepithelium in NMO cannot be precluded given expression of AQP-4 in this structure also.

Pathological studies point to early involvement of olfactory bulb and tract pathology in not only MS but also NMO and ADEM.11 MS and NMO cases, which came to autopsy within only months of disease onset, demonstrate striking olfactory bulb and tract pathology suggesting early involvement of these structures.11 Hyperacute involvement of olfactory bulb and tract structures is supported by extensive inflammatory pathology in ADEM cases, which died within days of symptom onset.11 As in MS, upper respiratory tract involvement has been linked to later emergence of ADEM,68 once again possibly linking involvement of neuroepithelial structures with olfactory bulb and tract. While olfactory assessments in patients with ADEM are impractical given the reduced level of consciousness that heralds diagnosis of this condition, we would predict prodromal dysfunction of olfactory threshold, with early impairment in olfactory discrimination and identification based on these pathological observations. Further, the association between olfactory bulb/tract and adjacent cortical demyelination provides reason to consider that the olfactory system may provide a conduit for the initiation and subsequent propagation of these diseases. It is interesting to note that the olfactory bulb/tract are juxtaposed to inferior frontal and temporal cortices, which are among the commonly affected cortical regions in MS (see figure 5).69 This would support the evolution of olfactory dysfunction that occurs throughout the course of MS where this has been extensively studied.

Figure 5

Schematic of the ‘nose to brain hypothesis’ and relationship to olfactory demyelination. Illustration of the ‘nose to brain’ hypothesis. An environmental agent (eg, virus, bacteria, cigarette smoke; red particles) infiltrates the nasal passage and sinuses. In a genetically susceptible individual, this could establishes an inflammatory milieu in the olfactory bulb (marked by *), a structure in direct contact with the sinuses through the fenestrations in the cribriform plate. Microglial (blue star-like structures) and lymphocytic (indigo circles) activation leads to demyelination (orange) in the olfactory bulb and tract, which relates to cortical demyelination, possibly via meningeal inflammation (arrow). The establishment of meningeal inflammation in this region could generalise to other central nervous system areas leading to more diffuse pathology typically encountered in the disease.11 ,71 ,72

The nature of olfactory loss changes later in the disease course where olfactory identification and discrimination impairments becoming predominant features in established RR and progressive disease.19 ,41 ,42 ,44 The anatomical basis for problems in olfactory discrimination and identification is complex as both olfactory bulb/tract and cortex have been implicated in these olfactory domains. In comparison to olfactory threshold abnormalities which appear restricted to the neuroepithelial compartment, the involvement of olfactory bulb/tract and cortex in RR and progressive MS could merely reflect more diffusely distributed pathology as is recognised with disease evolution. This might also explain why deficits in olfaction in MS have been linked to the extent of disability (as measured by the ambulatory-biased EDSS).19 ,31 ,35 ,37 ,44 ,46 ,51 ,52 While the degree to which olfactory defects in MS relate to cognitive dysfunction, and depression remains an area of controversy,31 ,40 ,51 ,52 intriguingly, a contribution of olfactory bulb/tract pathology to cognitive problems and depressive symptoms warrants consideration based on mouse models where olfactory bulbectomy drives depressive behaviour and global cognitive dysfunction.70

Implications for disease pathogenesis

Does pathological involvement of the olfactory bulb and tract in MS, ADEM and NMO have implications for the pathogenesis of these diseases? Several observations suggest that this may be so. The olfactory system connects the external world (olfactory receptor cells in the nasal mucosa) to the CNS (olfactory bulb/tract and subjacent meninges and cortex). Animal models have shown that olfactory structures are vulnerable to invasion by diverse pathogens, some of which can bypass the blood-brain barrier to penetrate deep regions of the brain and induce an inflammatory response to cause disease.71–73 This has led some to hypothesise that environmental agents enter the brain via this route to cause and/or propagate diseases where anosmia is an early feature, such as Parkinson's and Alzheimer's (ie, ‘olfactory vector’/‘nose to brain hypothesis’).11 ,71 ,72 As anosmia can similarly occur early in MS, pathological involvement of the olfactory bulb/tract and its relationship to adjacent cortical demyelination may not only explain this symptom but also may spin a different light onto observations which link chronic sinusitis to MS susceptibility, relapse risk and its demography (figure 5).64 ,65 In ADEM where olfactory bulb/tract demyelination is highly inflammatory and symptom onset occurs within days, disease is typically preceded by contraction of an upper respiratory tract infection (URTI).68 Might the olfactory system play a role in permitting a nasopharyngeal agent into the CNS where it initiates a process of acute or chronic neuroinflammation? While such questions arise from evidence that is only circumstantial at this point in time, they highlight the importance of further investigation into the olfactory manifestations of CNS demyelinating diseases and suggest that the olfactory demyelination observed in these conditions might have implications, which extend far beyond loss of smell alone.


The olfactory system is complex and functionally interdependent, comprising peripherally the nasal apparatus and the olfactory neuroepithelium, and centrally the olfactory bulb/tract and the olfactory cortex, situated in the inferofrontal and mesial temporal cortices. The conflation of clinical, psychophysical, electrophysiological, radiological and pathological data suggest that the olfactory system is commonly affected in CNS demyelinating diseases and can have a significant impact on quality of life. The fact that olfactory pathology in CNS demyelinating diseases is frequent, can occur early, be highly inflammatory and relates to pathology of closely juxtaposed cortex raises intriguing questions about the role of the olfactory system in the initiation and propagation of these diseases. Further research focused on deciphering the nature, extent, and underlying mechanisms of olfactory system involvement in MS, NMO and ADEM could shed critical pathogenic insight into these enigmatic diseases.


The authors would like to acknowledge Professor Margaret Esiri for kindly offering to proof read the manuscript and Aimee Taylor, medical illustrator, for her assistance in creating figure 5 entitled Illustration of the ‘Nose to Brain Hypothesis’.



  • Contributors AJ and GCD jointly conceived of the idea for this review article. AJ carried out a review of literature as outlined in text and put together the figures. AJ and GCD together interpreted the literature and wrote the manuscript. Both gave final approval for the submitted article and take joint responsibility for the accuracy and integrity of this work.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.