How to Diagnose Frontotemporal Dementia: Key Tests

Diagnosing frontotemporal dementia relies on a layered process that combines clinical observation, structured behavioral questionnaires, brain imaging, fluid biomarkers, and sometimes genetic testing. There is no single blood draw or scan that confirms it. Instead, clinicians piece together evidence from multiple angles, often over months or years. About half of people eventually diagnosed with the behavioral variant of FTD first receive a psychiatric diagnosis, and average diagnostic delay runs five to six years from symptom onset, making FTD one of the most frequently misidentified forms of dementia.

The Revised Clinical Criteria and What They Look For

The most widely used framework for diagnosing the behavioral variant of FTD (bvFTD) was published by an international consortium in 2011. Under these criteria, a “possible” diagnosis requires that a person show at least three of six core features: behavioral disinhibition, apathy or inertia, loss of sympathy or empathy, repetitive or compulsive behaviors, hyperorality or dietary changes, and a pattern of executive-function problems on neuropsychological testing.¹ A “probable” diagnosis adds two further requirements: the person must have clear functional decline, and brain imaging must show the pattern of atrophy or reduced metabolic activity expected in FTD. A definite diagnosis, in formal terms, requires either autopsy confirmation of the underlying brain pathology or identification of a known disease-causing genetic mutation.

These 2011 criteria were a meaningful upgrade over an earlier 1998 framework. When the two sets of criteria were compared head-to-head in the same patient group, all patients met the newer criteria, but only about two-thirds met the older ones. The discrepancy arose because the 1998 rules excluded patients who had spatial disorientation or early severe memory loss, features that do sometimes show up in FTD, especially in people carrying the C9orf72 gene expansion.² In validation studies, the revised “possible” criteria achieved a sensitivity of about 95% and a specificity of 82%, while the stricter “probable” criteria traded some sensitivity (around 85%) for higher specificity (95%).³

Language Variants and How They Are Classified

FTD is not just one disease. Alongside bvFTD, there is a family of language-dominant presentations grouped under the umbrella of primary progressive aphasia (PPA). An international working group established criteria for three main PPA subtypes: the nonfluent/agrammatic variant, where speech becomes halting and grammar breaks down; the semantic variant, where people gradually lose the meaning of words and concepts; and the logopenic variant, where word-finding pauses dominate but grammar stays relatively intact.⁴ A person is first diagnosed with PPA broadly and then classified into a subtype based on the specific speech and language features present. If imaging shows the expected pattern of brain atrophy for that subtype, the classification is upgraded to “imaging-supported.”

In practice, not everyone fits neatly into one of the three boxes. Data-driven analyses of language test results broadly confirm that these clusters exist, but somewhere between 15% and 30% of PPA patients remain unclassified. A fourth “mixed” subtype has been proposed to cover those who straddle categories.⁵ For a clinician working through a diagnosis, this ambiguity means that a clean label may not always come quickly, and serial assessments over time can be more revealing than a single snapshot.

Behavioral Questionnaires and Cognitive Testing

Standard cognitive screening tools, the kind used in routine memory clinics, often miss early bvFTD. That is because bvFTD initially attacks personality, social behavior, and executive function rather than raw memory or orientation, the domains those screens are designed to catch. Research on high-functioning bvFTD patients found that they performed normally on most traditional neuropsychological tests but showed clear impairments on a battery specifically designed to probe executive and social cognition.⁶ Tests of facial emotion recognition and the ability to detect social faux pas are especially useful, since these abilities deteriorate early in genetic forms of FTD and can even show deficits before symptoms become obvious.⁷

Informant-based questionnaires that ask a family member or caregiver about behavioral changes add another layer. A recently validated tool called the Behavioral Evaluation Scale of FTD (BES-FTD) distinguished bvFTD from both healthy controls and Alzheimer’s disease with a specificity over 93%, outperforming an older questionnaire used for the same purpose.⁸ The Behavioural Dysfunction Questionnaire (BDQ) has also shown promise in separating bvFTD from Alzheimer’s disease and major depression, though researchers found that informants’ recall of when specific symptoms began was unreliable, often shifting by more than a year between assessments.⁹ That finding is a practical reminder: pinning down exact symptom timelines from family members is harder than it sounds, and clinicians who rely too heavily on recalled dates may be working with shaky data.

Brain Imaging

Structural MRI is often the first imaging step. In bvFTD, the earliest atrophy tends to appear in structures deep in the brain, including the amygdala, the striatum, and the insula, before spreading to the frontal and temporal cortex. In the semantic variant of PPA, the amygdala also shrinks early, followed by the left temporal lobe and hippocampus. In the nonfluent variant, atrophy starts in the left striatum, insula, and thalamus before reaching the opercular regions involved in speech production.¹⁰ These staging patterns can help clinicians match what they see on a scan to what they observe at the bedside. One MRI-based biomarker approach compares the volume of frontal and temporal regions to posterior regions, capitalizing on the fact that FTD preferentially attacks the front of the brain while the back is relatively spared.¹¹

FDG-PET, which maps how actively brain regions are consuming glucose, offers better sensitivity and specificity than MRI alone. The best diagnostic accuracy comes from using both techniques together.¹² A distinctive feature seen on FDG-PET in FTD is asymmetric reduced metabolism, more commonly lateralized to the left hemisphere, which helps differentiate it from Alzheimer’s disease.¹³ Amyloid-PET, which detects the amyloid plaques characteristic of Alzheimer’s, serves mainly as a rule-out: a negative amyloid scan in someone with dementia symptoms nudges the diagnosis away from Alzheimer’s and toward FTD or another non-amyloid condition.¹⁴ Tau-PET exists but remains largely confined to research settings for now.

Fluid Biomarkers in Blood and Spinal Fluid

Cerebrospinal fluid (CSF) analysis can be valuable, not so much for confirming FTD directly as for ruling out Alzheimer’s disease. In FTD, CSF levels of total tau and beta-amyloid tend to be in the normal range, whereas in Alzheimer’s, tau shoots up and beta-amyloid drops.¹⁵ The ratio of total tau to beta-amyloid in CSF is especially useful and performs better at separating FTD from Alzheimer’s than either marker alone.¹⁶

Neurofilament light chain (NfL) is a different kind of marker. It reflects axonal damage, and it rises in FTD. A meta-analysis found that NfL can help differentiate FTD patients from healthy people, from people with Alzheimer’s, and from those with psychiatric disorders.¹⁷ Conveniently, NfL can now be measured in a blood draw rather than requiring a spinal tap. Plasma NfL levels are highest in people who have FTD with co-occurring motor neuron disease (the fastest-progressing form), followed by the semantic and behavioral variants. Plasma NfL is also significantly higher in true FTD than in so-called “phenocopy” cases, patients who look like bvFTD clinically but do not actually have progressive neurodegeneration.¹⁸ That distinction matters because phenocopy patients do not worsen over time and have a fundamentally different prognosis.

Genetic Testing

Around 30% of people with FTD have a strong family history of the disease.¹⁹ The three genes that account for most inherited cases are C9orf72 (a repeat expansion on chromosome 9), GRN (progranulin), and MAPT (the tau gene). In one large North American series, about 5% of even apparently sporadic (non-familial) FTD patients carried mutations in one of these genes.²⁰ The frequency of each gene varies by geography. In a Serbian cohort, for example, C9orf72 expansions were found in about 7% of FTD cases overall, but no MAPT mutations were detected at all, suggesting regional variation in which mutations predominate.²¹

In a clinical context, genetic testing is most clearly indicated when there is a family history of dementia or motor neuron disease. But the fact that mutations occasionally appear in people with no known family history means testing can sometimes be informative even in sporadic cases. The results can also help predict the type of underlying brain pathology, since each gene maps to a specific protein abnormality. Rarer genes like TARDBP, PSEN1, and VCP occasionally turn up as well, so broader gene panels are increasingly offered.²⁰

Why FTD Is So Often Misdiagnosed

The behavioral variant of FTD is a diagnostic minefield because its early symptoms, including apathy, impulsivity, social inappropriateness, and emotional flatness, overlap heavily with psychiatric conditions. Roughly half of bvFTD patients receive a psychiatric diagnosis first, and the average delay from symptom onset to the correct diagnosis stretches to five or six years.²² In one case series, about 69% of FTD patients had previously been diagnosed with another mental disorder, and among those, 71% had been labeled with at least moderate depression.²³ People initially diagnosed with psychotic depression waited the longest for a correct FTD diagnosis, averaging over four years of diagnostic lag.

Even brain imaging can be misleading. Depression itself causes reduced metabolic activity in the frontal lobes on FDG-PET, the same pattern associated with FTD. PET’s sensitivity for frontal dementia is over 90%, but its specificity is lower, in the range of 68–75%, leaving room for false positives.²⁴ A depressed patient whose PET scan shows frontal hypometabolism can easily be misdiagnosed with FTD, and the reverse mistake, FTD mislabeled as treatment-resistant depression, is just as common. The diagnostic path through this ambiguity typically requires serial assessments over time, tracking whether symptoms respond to psychiatric treatment (suggesting depression) or progressively worsen despite it (suggesting neurodegeneration).

There are also sex-related pitfalls. A Bayesian analysis found that men with bvFTD experienced diagnostic delays roughly two years longer than women.²⁵ Conversely, women with bvFTD were more likely to have received a prior depression diagnosis before being correctly identified, and they presented with lower cognitive test scores at the time of FTD diagnosis.²⁶ These patterns suggest that the diagnostic process may be influenced by gendered assumptions about behavior: a man who becomes socially withdrawn may be written off as “just how men are” for longer, while a woman exhibiting the same changes may be steered toward a depression diagnosis.

The Neurological Exam and Motor Neuron Signs

A standard neurological examination is part of the diagnostic workup, and it can surface findings that most people do not associate with dementia. FTD and amyotrophic lateral sclerosis (ALS) are now understood to sit on a shared disease spectrum. Some patients diagnosed primarily with FTD develop signs of motor neuron disease such as muscle wasting, fasciculations, or abnormal reflexes.²⁷ Finding these signs changes the clinical picture substantially, both because FTD with motor neuron disease progresses faster and because the overlap strengthens the diagnostic confidence (the combination of behavioral changes plus motor neuron signs is fairly distinctive). Clinicians typically check for primitive reflexes, muscle tone changes, and signs of upper and lower motor neuron involvement during the exam.

Emerging Diagnostic Tools

Two newer areas of research may eventually change how FTD is detected, though neither has entered routine clinical use yet. The first is automated speech analysis. In a study of over 200 participants, a computational system that extracted features from recorded speech samples distinguished between the three PPA variants with strong accuracy, achieving an area under the curve of 0.90 in both the original and an independent validation cohort.²⁸ Separate work using natural language processing on short picture-description tasks found that the balance of content words versus function words, along with the use of elaborative elements like adjectives and adverbs, could help distinguish PPA subtypes.²⁹ The appeal of these approaches is that a brief speech sample could one day serve as a low-cost screening tool in a general practitioner’s office.

The second emerging frontier is retinal imaging using optical coherence tomography (OCT), a quick, noninvasive eye scan. FTD patients show thinning of the outer retina compared to healthy controls, a finding that differs from Alzheimer’s, where inner retinal layers are more affected.³⁰ The outer retinal thinning in FTD correlated with scores on a standard cognitive screening test, suggesting it tracks disease severity.³¹ Retinal nerve fiber layer thinning has also been documented in FTD, and the degree of thinning was more pronounced than in mild Alzheimer’s disease.³² OCT is already widely available in ophthalmology offices, so if these findings hold up in larger studies, it could become an accessible and inexpensive addition to the diagnostic toolkit.

What Happens Under the Microscope

Although autopsy is the only way to definitively confirm the specific type of brain pathology behind FTD, understanding the pathological landscape helps explain why diagnoses can shift during life. Most FTD cases are caused by abnormal accumulations of either TDP-43 or tau protein. A third, rarer protein called FUS accounts for the majority of cases that are negative for both TDP-43 and tau.³³ FUS pathology makes up roughly 10% of FTD cases overall and tends to present with a distinctive clinical picture.³⁴ The fact that both FTD and ALS can be driven by TDP-43 or FUS reinforces their biological relationship and helps explain why motor neuron signs sometimes appear in patients diagnosed with FTD.³⁵

This pathological complexity is one reason that clinicians think of FTD diagnosis as an evolving process rather than a single event. Someone initially classified with probable bvFTD based on behavior and imaging may later be found, through genetic testing, to carry a C9orf72 expansion, which predicts TDP-43 pathology. Or a person diagnosed with the semantic variant of PPA may develop motor neuron signs years later, shifting the clinical picture. The diagnostic framework is designed to accommodate these revisions, moving from “possible” through “probable” and, when evidence accumulates, toward pathological certainty.