June 2026 Research Update

This group of papers points to a common goal: improving how well progressive supranuclear palsy (PSP) can be distinguished from look-alike conditions such as Parkinson's disease (PD), multiple system atrophy (MSA), and corticobasal syndrome (CBS). Overall, MRI-based measurements had the strongest and most consistent accuracy across large groups, while blood-based metabolite patterns, diffusion MRI tract mapping, and video-based facial analysis were earlier-stage tools that still need broader testing. [1] [2] [3] [4]

In a very large, multi-cohort study (2,111 people), a simple MRI measurement called the dual-line midbrain PSP index (DMPI) separated PSP from non-PSP parkinsonism with a high area-under-the-curve of about 0.97, and it had fewer "gray-zone" results than other MRI markers. The same MRI approach also performed well in early-stage disease (AUC 0.97) and in a subgroup with diagnoses confirmed by pathology (AUC 0.94), which supports its potential reliability across settings. [1]

Another MRI-focused study looked at a different real-world diagnostic confusion: enlarged brain ventricles that can mimic normal pressure hydrocephalus on imaging. In this retrospective study of 125 movement-disorder patients, people with PSP had a higher Evans Index (a ventricle size measure), and the Evans Index exceeded 0.3 only in PSP, but a typical hydrocephalus pattern (DESH) was absent in all groups and the callosal angle stayed above 90 degrees across diagnoses. The authors concluded that ventricle size alone can be misleading in PSP, and that looking for other imaging features may help avoid confusing PSP-related changes with hydrocephalus patterns. [5]

Several studies explored nontraditional ways to support diagnosis and subtyping, with results that were encouraging but still preliminary. A prospective single-center pilot study used blood-based metabolite profiles and machine learning to separate PD from atypical parkinsonism (including PSP and MSA) and to separate PSP from MSA, but the authors emphasized the need for multicenter validation because subgroup sizes were limited. [2]

Two imaging and video studies focused on patterns that might reflect where disease affects the brain and body. A diffusion MRI study mapped tract-level "degeneration patterns" and reported that PSP involved specific pathways such as the dentato-rubro-thalamic tract and superior cerebellar peduncle, enabling group differentiation and single-patient visualization, but external validation and test-retest reliability were not yet established. Separately, a small diagnostic study (62 participants) analyzed short facial videos and found that a machine-learning model could differentiate PSP from PD with 0.82 accuracy in a held-out set, using features like mouth-opening speed and forehead-wrinkle variability, though performance still needs confirmation in broader populations. [3] [4]

In people diagnosed with corticobasal syndrome (CBS), a multimodal biomarker study showed that brain metabolism patterns on a PET scan could sort patients into subgroups that aligned with different underlying protein pathologies, including a PSP-like subgroup. In 113 CBS patients, the PET metabolic patterns showed high specificity for a PSP tau profile (96.2%) and high sensitivity for a corticobasal degeneration tau profile (97.7%), supporting the idea that metabolic imaging patterns can assist a biomarker-based workup when the clinical syndrome alone is not specific. [6]

Finally, a review discussed the "procerus sign" (a facial expression pattern sometimes described in PSP) as a clinical clue rather than a confirmed test. The authors noted it is commonly observed and may appear early, but its diagnostic value is uncertain, it is not part of current PSP diagnostic criteria, and better prevalence studies are needed. [7]

Right now, the most mature direction in this set of studies is using MRI measurements to support the clinical diagnosis when symptoms overlap with other conditions, and to reduce ambiguity in borderline cases. Blood-based metabolite testing, advanced diffusion MRI mapping, and video-based facial analysis are mainly background research at this stage: they suggest future ways to make diagnosis more objective or earlier, but they need replication in larger and more diverse groups (and across multiple clinics) before their accuracy and practical role are clear. Clinical signs described in reviews, like the procerus sign, may contribute to diagnostic suspicion but are not proven enough to stand alone as a diagnostic test. [1] [5] [2] [3] [4] [6] [7]

Across these studies, the overall pattern was that progressive supranuclear palsy (PSP) commonly affects executive function (planning, flexibility, and problem-solving) and processing speed, with additional changes in visual-spatial skills and verbal fluency. Some areas, such as naming and certain memory measures, appeared relatively more preserved, although results varied by test and by PSP subtype. A recurring issue was that motor and eye-movement problems can partly influence performance on some cognitive tests, so test results may reflect a mix of thinking changes and physical limitations. [8] [9] [10]

In a cross-sectional study of 34 adults with PSP in Lima, Peru, cognitive testing showed broad difficulties, with especially clear problems in processing speed, executive function, and episodic memory, while naming was relatively preserved. The study also reported differences across clinical subtypes: PSP-Richardson syndrome showed the greatest cognitive difficulties, PSP with progressive gait freezing looked more preserved, and PSP-parkinsonism showed a specific reduction in delayed episodic memory. Initial symptoms (such as whether parkinsonism was the first symptom) did not strongly predict overall cognitive level in this group. [8]

In 81 people with PSP-Richardson syndrome, performance on the Trail Making Test (a timed test often used to assess flexible thinking and attention) was mainly explained by executive-function problems. Measures of movement severity and eye-movement impairment did not significantly explain Trail Making Test performance in their statistical model, suggesting this test largely reflects thinking changes rather than physical slowing in this specific setting. [9]

A larger analysis pooled cognitive data from two phase 2 clinical trials (PASSPORT and tilavonemab) and followed participants for up to 52 weeks, which allowed a clearer look at which abilities change fastest over time. The greatest declines were seen in visual-spatial tasks, executive-type tasks, and semantic fluency, while immediate and delayed memory changed slowly or sometimes not at all. When the researchers statistically adjusted for motor and eye-movement impairment, the amount of measured cognitive decline became smaller for several tests, suggesting that worsening movement and eye control can partly reduce cognitive test scores even when the main issue being measured is thinking. [10]

For patients and caregivers, these studies support that thinking changes in PSP are often most noticeable in planning, switching between tasks, keeping up speed, and handling visual-spatial information, and that these areas may change over time. They also highlight an important uncertainty in real-world interpretation of testing: some cognitive scores can drop partly because of PSP-related movement and eye-movement limitations, depending on the specific test. This body of work is mainly about understanding patterns of cognitive change and improving how cognitive tests are interpreted in PSP, rather than changing care right now. [8] [9] [10]

Across these papers, PSP progression looks like a gradual build-up of disability over several years, with serious events (like aspiration pneumonia and fractures) appearing later in the course. The research also suggests that measurable brain changes on imaging increase over time and may track how quickly symptoms are changing, but this imaging work is still exploratory. [11] [12]

In a single-center Japanese review of 68 people with probable PSP, the time from first symptoms to first specialist visit was about 25 months, and the time to diagnosis was about 33 months, showing a substantial diagnostic delay. The study also estimated later milestones from symptom onset, including admission for aspiration pneumonia at about 59 months and death at about 62 months. The authors cautioned that comparing survival between people diagnosed earlier vs later can be misleading depending on whether survival is counted from symptom onset or from diagnosis, so the delay itself should not be treated as the cause of survival differences. [11]

A longitudinal brain scan study using a PET measure of brain energy use identified pattern changes linked to different PSP types and followed how these patterns changed over time. Pattern levels generally increased at follow-up and rose alongside worsening clinical status, suggesting they may track progression. Higher baseline pattern levels also predicted which people with a variant form later converted to the classic PSP-Richardson type, and one pattern (linked to progressive gait freezing) showed more involvement of deep movement-related brain areas. [12]

For patients and caregivers, these studies help set expectations about the pace of PSP and the kinds of complications that tend to appear later, while also explaining why getting a clear diagnosis can take years. The imaging results matter mainly as background research for now: they suggest a possible way to measure progression and identify who may shift into the classic PSP pattern, but this approach is not yet a standard part of care and needs further confirmation.

Across two small, early-stage studies, rehabilitation was described as focusing on practical goals like transfers, walking stability, and daily activities. The overall direction of results was improved balance and functioning in the short term, while it remains unclear how much these gains last or how well they apply to all people with progressive supranuclear palsy (PSP). [13] [14]

In one individual with PSP with cerebellar predominance (a form where coordination and gait can be strongly affected), a comprehensive physiotherapy program emphasized transfer training, building independence with activities of daily living, and addressing stiffness and tremor. The report highlighted that diagnosis can be challenging and described potential functional improvement despite limited disease-specific treatment options. [13]

In a pilot case series of 7 adults with extrapyramidal disorders (including PSP), immersive virtual reality training using a body weight support system was added to usual inpatient physical therapy for 10 days. Balance scores improved (for example, the Berg Balance Scale increased from 32.4 to 35.7), while walking speed changes did not reach statistical significance. No serious adverse events were reported; mild, short-lived symptoms like fatigue, eyestrain, and dizziness were noted, and participants reported high satisfaction. [14]

For patients and caregivers, these studies add background evidence that rehab approaches in PSP may reasonably prioritize safety, balance, transfers, and independence with daily activities, and that technology-assisted balance training can be feasible in supervised inpatient settings. Because one paper describes a single person and the other is a very small uncontrolled study with short follow-up, it is still uncertain how reliably these improvements apply to different PSP subtypes, outpatient settings, or longer time periods. [13] [14]

This month, several studies used donated brain tissue to describe what progressive supranuclear palsy (PSP) can look like under the microscope and how it can overlap with other conditions. Overall, the papers point to PSP having distinct tau patterns (often involving non-neuron brain cells), frequent mixed pathology with other diseases in some movement-disorder donors, and early signals that brain energy (mitochondria-related) proteins may differ in PSP. [15] [16] [17]

In a postmortem study of people whose clinical symptoms looked like primary progressive aphasia (language-led symptoms) or behavioral-variant frontotemporal dementia (behavior-led symptoms), PSP brains tended to show more tau in glial (support) cells, while corticobasal degeneration tended to show more tau in neurons. The study also reported that language-led cases had more left-sided tau, with the anterior temporal lobe most involved, while behavior-led cases were more symmetric or right-sided, with the middle frontal gyrus as a peak region. [15]

In a large brain-bank analysis across movement disorders, clinical diagnoses did not always match what was found at autopsy, with misdiagnosis rates around 10-20% across movement disorders. The same study reported that co-existing brain diseases were common in some groups, including Alzheimer-type changes occurring in 40.0% of Lewy body disease cases, and it found that pathological diagnoses differed by ancestry, with South Asian donors more likely to have PSP in this brain-bank sample. [16]

In a small, exploratory protein study using postmortem brain tissue (4 PSP donors and 4 controls), researchers highlighted a set of six proteins that formed a connected network related to mitochondria (the cell's energy system). Follow-up staining suggested two of these proteins were present in neurons, and one (PPA1) also showed prominent signal in microglia in PSP brains, though the authors emphasized the small sample size and that bulk tissue mixes many cell types. [17]

Separately, one laboratory report described creating an induced pluripotent stem cell (iPSC) line from the blood of a single person with PSP (Richardson syndrome) of Indian ethnicity. This type of cell line is mainly a research tool that can be used to study disease processes in lab-grown cells. [18]

For patients and caregivers, these studies mainly add background about why PSP can be hard to classify and why symptoms can vary: brain changes can differ by symptom pattern, and mixed brain pathology can occur in some people. The mitochondria-related protein results and the new stem cell line are early-stage research tools rather than tests used in care today, and it is still uncertain how directly these lab signals connect to day-to-day symptoms or to future diagnostics. [15] [16] [17] [18]

Across three case reports, the main pattern was that progressive supranuclear palsy (PSP) did not begin with the classic set of symptoms people often hear about. Instead, PSP appeared alongside systemic symptoms, long-standing speech problems, or nerve and muscle-type signs that could point clinicians toward other diagnoses first. [19] [20] [21]

In one 83-year-old woman, worsening walking problems and overall decline began after a mild SARS-CoV-2 infection, and PSP was later considered likely based on exam features plus supportive brain imaging. A trial of levodopa did not improve symptoms in this patient, showing how limited medication response can add to diagnostic uncertainty when the early picture is not typical. [19]

In another single patient followed over time, a speech-first condition (primary progressive apraxia of speech) later developed PSP-like movement symptoms and a specific brainstem imaging change called hypertrophic olivary degeneration. Autopsy confirmed PSP-type brain changes, and the report suggested this brainstem change can appear later in the illness even without classic symptoms often linked to it (such as palatal tremor). [20]

A third case showed that PSP can rarely look like amyotrophic lateral sclerosis (ALS) at first, with visible tongue fasciculations (twitching) and electromyography (EMG) signs of denervation. In that patient, PSP was confirmed at autopsy, underlining that similar outward symptoms can come from different underlying brain diseases. [21]

For patients and caregivers, these reports are mainly background information about why PSP can be hard to recognize early and why doctors may consider (or rule out) other conditions when symptoms do not follow a typical pattern. Because each paper describes only one person, it is not clear how often these unusual presentations happen or whether the same sequence of symptoms would occur in others. [19] [20] [21]