Headache-Associated Phantosmia as a Harbinger of Lewy Body Dementia

A 77-year-old left-handed man with the factor V Leiden mutation, hyperlipidemia, and sensorineural hearing loss presented to a headache clinic with dull and vise-like daily headaches that were typically followed by smelling a foul odor within the hour. The headache was predominantly located posteriorly, with intermittent frontal involvement and extension above both ears. He was often awakened by the headache, which would then lessen during daytime. During an episode, he would appear distraught but not confused.

He reported headache episodes without olfactory symptoms starting 2 years ago, and onset of accompanying smell in the prior 14 months. He was initially managed by an otolaryngologist for abnormal smell sensation. He also reported recurrent falls starting about 5 years prior. His initial fall was with head strike and loss of consciousness, for which he did not seek medical care, but his ambulation had improved over the ensuing 18 months and he did not experience additional falls.

His evaluation prompted concern for phantosmia, likely representing the aura of a primary headache disorder versus epilepsy. Posttraumatic headache was also considered, although the temporal correlation of his falls to his initial headaches was not consistent given that headache onset occurred years after his fall with traumatic brain injury (TBI). Lamotrigine was initiated as a preventative treatment for headache and for the potentially underlying epilepsy. The plan of care included admission to the epilepsy monitoring unit (EMU) for spell capture and, after his family reported cognitive changes, referral to the behavioral neurology clinic (BNC).

In the BNC, the patient reported worsening short-term memory, difficulty with name recall, slowed responses, and vision changes. His family noticed additional word-finding problems, worsening handwriting and spelling, and misplacement of items. These changes were present for at least 3 years, resulting in the inability to manage his business, but had worsened significantly over the last 6 months. Multiple mistakes were made in personal finances, including inappropriately sharing sensitive information. His family expressed increasing concerns about his ability to drive safely, noting out-of-character incidents of aggressive driving and attempts to take the wheel as a passenger. The patient recently misidentified a stranger as a close relative and misinterpreted his hotel as a clinic.

Notably, behavior changes were observed prior to his falls, headaches, and cognitive changes. New marital discord started about 6 years before his evaluation, driven by his suspicions that his wife was unfaithful. He later became increasingly concerned about strangers trespassing on his property, leading to him regularly carrying a firearm.

Current medications included those for vascular risk factors and lamotrigine (75 mg, twice daily). He was a remote former smoker, but he had no recent alcohol use and no prior illicit drug use. Family history was unremarkable. His neurologic examination disclosed mild parkinsonism, including bradykinesia with decrement and a slowed, stooped walk with reduced arm swing, but without tremor. Cognitive evaluation with the Montreal Cognitive Assessment yielded a score of 14/30 (higher scores indicate better performance), with the most profound deficits in visuospatial/executive function, language, abstraction, delayed recall, and orientation.

Serum tests for reversible causes of dementia (e.g., thyroid stimulating hormone, rapid plasma reagin, vitamin B₁₂) were unremarkable. Brain magnetic resonance imaging (MRI) performed a year prior showed old left lentiform nucleus and external capsule lacunes, prominent vascular spaces in the basal ganglia, and atrophy with predominance in the posterior regions. A routine electroencephalogram disclosed no epileptiform activity or other abnormalities. His EMU stay revealed two events of mild phantosmia without electrographic correlates.

The patient met clinical criteria for dementia given his progressive cognitive decline that interfered with daily activities. The additional clinical history and examination were most concerning for LBD and, more specifically, dementia with Lewy bodies (DLB). Despite multiple vascular risk factors, his imaging was not consistent with vascular dementia. The differential diagnosis also included cognitive impairment due to ongoing subclinical seizures, because an unrevealing EMU stay does not conclusively rule out epilepsy. Other neurodegenerative disorders were considered, but the patient did not clearly meet the clinical criteria for behavioral variant frontotemporal dementia, despite the early marital discord (1), and there were no other features to suggest other Parkinson-plus syndromes such as supranuclear palsy (2) or corticobasal syndrome (3). He initially met criteria for probable Alzheimer’s disease, except for the exclusionary criterion of DLB core features (4).

The DLB clinical criteria require dementia (5), the initial deficits of which most notably manifest in altered executive function, visuospatial abilities, attention/processing speed, and (in some cases) language and memory (6). Core clinical features include prominent cognitive fluctuations, well-formed visual hallucinations (VHs), rapid eye movement sleep behavior disorder, and parkinsonism (Table 1) (5). DLB is distinguished from Parkinson’s disease with dementia (PDD) by the “1-year rule”: a diagnosis of PDD must include a clear Parkinson’s disease diagnosis for at least 1 year before the onset of dementia (5). Notably, LBD is an umbrella term that includes both DLB and PDD. An unusual feature of the patient described in this case is the appearance of early paranoid delusions about infidelity, because delusions typically arise later in the course of LBD (6).

The most novel feature of the patient described in this case is that the inaugural and most prominent hallucination was olfactory. One year after he was first seen in the BNC, he described a well-formed VH of “gentlemen” sharing his room whom he could physically guide with touch, consistent with tactile hallucinations previously described in DLB (7). Given his dementia, cognitive fluctuations, VHs, and parkinsonism, he met the clinical criteria for probable DLB without biomarkers (Table 1) (5). Over time, the patient’s Lewy Body Composite Risk Score (LBCRS; range 0–10) (8) was five (Table 2). An LBCRS of three or more indicates probable DLB and assists in distinguishing DLB from Alzheimer’s disease with a sensitivity of 94.2% and a specificity of 78.2%, and from any other dementia with a sensitivity of 97.9% and a specificity of 86.1% (8). His TBI was thought to be unrelated to his later diagnosis of DLB, because the current literature does not support a strong association (9–12), but his falls likely represented unrecognized parkinsonism during his early disease course.

His headaches were initially prominent despite up-titration of lamotrigine but became less frequent and then ceased over months, at which time lamotrigine was titrated off without headache recurrence. His phantosmia remained persistent. He also developed features consistent with olfactory reference syndrome (ORS), which is categorized under other specific obsessive-compulsive and related disorders (13); he would wake up perceiving a foul odor emanating from his own body that prompted nightly showers. ORS is rare, generally has an early age of onset (average=21 years), and is often associated with other comorbid psychiatric diagnoses (14). ORS was described previously in Parkinson’s disease (15), but the prevalence of ORS in LBD is unknown.

Olfaction requires a functional olfactory neuroepithelium, olfactory bulb, and olfactory nerve (16). The primary olfactory regions (i.e., anterior olfactory nucleus, tenia tecta, olfactory tubercle, entorhinal cortex, piriform cortex, and amygdala) and secondary olfactory regions (i.e., thalamus, hypothalamus, and orbitofrontal cortex), with additional connections to the hippocampus and insular cortex, are all required for the emotional response to and the perception, identification, discrimination, and modality-specific memory consolidation of olfactory inputs (16). Olfactory function can be localized to the left or right cerebral hemisphere on the basis of lesional and functional brain imaging studies, but some authors have hypothesized a dominant role for the right temporal lobe in light of patients with epilepsy who underwent unilateral temporal lobectomy (16).

OHs are a type of qualitative olfactory dysfunction and involve a perception of a smell in the absence of an odor stimulant (16). This phenomenon, also known as phantosmia, contrasts with parosmia (also called troposmia), another qualitative olfactory dysfunction that refers to a distorted smell perception in the presence of an odor stimulant (16, 17). Hyposmia (reduced smell) and anosmia (lost smell) are quantitative olfactory dysfunctions (16).

Many nonneurological and neurological conditions are associated with qualitative or quantitative olfactory dysfunction; however, less is known about qualitative olfactory dysfunction (16, 18). Phantosmia is associated with a wide range of conditions; therefore, the differential diagnosis for new-onset phantosmia is broad (Table 3) (19–25). Phantosmia has been described in primary headache disorders (26–28), focal epilepsy (16, 29), multiple sclerosis (30), Parkinson’s disease (16, 19, 31), and following TBI (19, 32). Phantosmia has also been described in multiple nonneurologic medical conditions, including psychiatric disorders (33–35), post upper respiratory tract infections (19), sinonasal disease (19, 36, 37), and gastroesophageal reflux disease (38). Phantosmia is also correlated with some prescription medications (38). Idiopathic phantosmia is common but does not clearly predict the development of serious health conditions or neurodegenerative disorders when assessed longitudinally (39).

For quantitative olfactory dysfunction (i.e., hyposmia and anosmia), standardized clinical assessments for smell identification can be made, but no standardized approach exists for the clinical assessment of qualitative olfactory dysfunction. A detailed clinical interview performed by an experienced clinician is the most frequent approach (17), but standardized questionnaires have been proposed (32, 37, 38). Given the absence of standardized assessments, phantosmia is predicted to be underassessed and underidentified, despite an estimated prevalence of 4.9% (40) to 6.5% (32) in older adults. However, only a small portion (11.1%) of individuals with phantosmia discuss alterations in smell with a medical provider (32). Furthermore, the current literature regarding the duration, quality (pleasant, unpleasant, neutral), and other features (e.g., vague vs. precise; intrinsic vs. extrinsic) of the perceived odors associated with qualitative olfactory dysfunction and possible specific associations with the various etiologies is limited to case reports and case series (26, 31, 34, 41). Phantosmia is associated with parosmia (40), female sex (32, 40), younger age (32), low socioeconomic status (32), vascular risk factors (40), fair or poor health status (32), and brain-derived neurotrophic factor allele status (40). Given the association of phantosmia with COVID-19 infections (42), phantosmia prevalence is likely to increase.

There are multiple proposed mechanisms for phantosmia that depend on the clinical context (Table 3) (26, 29), including both peripheral and central etiologies (43). Regarding phantosmia in synucleinopathies, olfactory bulb alpha-synuclein has high specificity and sensitivity for multiple Lewy body disorders, including LBD and PD (44), but there is actually increased dopaminergic signaling within the olfactory bulb and anterior olfactory nucleus in Parkinson’s disease (31). Thus, it has been proposed that hyperexcitable dopaminergic activity in the setting of olfactory pathway denervation may be responsible for phantosmia in Parkinson’s disease (31). However, reduced thalamocortical cholinergic signaling resulting in enhanced cortico-thalamocortical and cortico-cortical signaling has been implicated in VH pathogenesis in DLB (45); therefore, it is appealing to hypothesize that disrupted cholinergic signaling to the piriformis (29) may play a role in OHs accompanying DLB. Regardless of the neurotransmitter system involved, it has been proposed that incomplete loss of olfactory sensory neurons or the disproportionate loss of inhibitory neurons may be responsible for overactivity that leads to parosmia and phantosmia (17).

The type and frequency of hallucinations in DLB include the more common VHs, which are followed by auditory hallucinations, OHs, and finally tactile hallucinations; notably, OHs are present in only 6.6% of patients (46). OHs are rare in neurodegenerative disorders and are most often accompanied by other types of hallucinations or delusions (47). Isolated OHs have been described in a subset of patients with Parkinson’s disease without cognitive impairment (48). However, VHs more commonly occur first in Parkinson’s disease and are usually followed by nonvisual hallucinations over time (49). To our knowledge, there are no reported cases of isolated OHs in DLB. Additionally, there is no literature on the prevalence of OHs at the time of DLB diagnosis, suggesting that OH is either a rare or an underassessed symptom of early DLB.

If the inaugural phantosmia in our patient was secondary to DLB neurodegeneration, it is intriguing to speculate about why the olfactory system was affected before the visual and sensory systems. Differences in temporal symptom onset between Parkinson’s disease and DLB may be related to the differential and sequential anatomy affected by synuclein pathology (50). Specifically, it has been proposed that DLB synuclein pathology begins in the olfactory bulb and propagates to nearby limbic regions, followed by higher associative neocortices in a rostrocaudal fashion, as opposed to the caudorostral Braak staging model for Parkinson’s disease (50). With a rostrocaudal model in mind, the piriform cortex would be affected early in DLB (50). Given the well-known gradient of inhibitory interneurons conferring a smaller population of inhibitory interneurons in the anterior piriformis (51), the rostrocaudal spread of synuclein pathology may lead to an even more disproportionate loss of inhibitory neurons in the anterior piriformis, leading to overactivity and, ultimately, to phantosmia. Along this line, the later spread of synuclein to the thalamus (50) may be responsible for the later onset of VHs via alteration of thalamocortical connections to the temporo-occipital visual areas (45).

Given the unusual presentation of symptoms and wide range of differential diagnoses for phantosmia, there was some delay in diagnosis even after the patient received specialized care in the BNC. Incorrect initial diagnoses are frequently made among patients with LBD; some reports document an 18-month average delay in the diagnosis of this disease (52). However, recognizing OHs as a potential early symptom of DLB may lead to earlier diagnoses in future cases.

We acknowledge the patient and his family, who made this case report possible.