Progressive Supranuclear Palsy Diagnostic Criteria and Classification Across Three Decades of Change

1. Introduction

Progressive supranuclear palsy (PSP) is a rare adult-onset neurodegenerative disorder and one of the most common atypical parkinsonian syndromes. Originally described by Steele et al., PSP is characterized by varying combinations of postural instability, ocular motor dysfunction, akinetic-rigid parkinsonism, cognitive impairment, and bulbar symptoms [1]. Neuropathologically, PSP is classified as a primary 4R tauopathy, characterized by the accumulation of hyperphosphorylated tau protein in neurons and glial cells, leading to NFTs, coiled bodies, neuropil threads, and the characteristic tufted astrocytes described in pathologic diagnostic criteria [2]. The disease typically presents after 60 years of age and is associated with substantial morbidity, disability, and reduced quality of life due to progressive gait impairment, recurrent falls, dysphagia, and cognitive decline [3].

Accurate diagnosis of PSP remains challenging because of the frequent dissociation between clinical presentation and underlying pathology. The 1996 NINDS-SPSP criteria provided the first widely accepted clinical diagnostic framework and demonstrated excellent specificity for the classical Richardson syndrome phenotype [4]. However, subsequent clinicopathological studies revealed that many patients with pathologically confirmed PSP initially present with alternative phenotypes, including PSP-P, PSP-PGF, frontal behavioral syndromes, speech and language disorders, and corticobasal syndrome [5]. These observations highlighted the substantial phenotypic heterogeneity of PSP and explained the limited sensitivity of earlier diagnostic criteria, particularly during the initial stages of disease.

Recognition of this broader clinical spectrum led to the development of the vPSP framework by Respondek et al., which demonstrated that Richardson syndrome represents only one manifestation within a diverse phenotypic continuum [6]. These findings subsequently informed the 2017 MDS criteria, which incorporated multiple PSP phenotypes, four core functional domains, and graded levels of diagnostic certainty to improve early diagnostic sensitivity while maintaining acceptable specificity [7]. More recently, the MAX rules were introduced to standardize phenotype assignment in patients fulfilling multiple MDS-PSP categories [8]. This review summarizes the historical evolution of PSP diagnostic frameworks, examines their strengths and limitations, and discusses future directions toward biomarker-supported and pathology-informed diagnosis of PSP.

2. Methods

A narrative literature review was conducted to summarize the historical evolution of diagnostic frameworks for PSP. A comprehensive search of PubMed/Medline, Scopus, and Google Scholar was performed from database inception through September 1, 2026. Search terms included combinations of “progressive supranuclear palsy,” “PSP,” “diagnostic criteria,” “classification,” “neuropathologic criteria,” “NINDS-SPSP,” “Movement Disorder Society criteria,” “MDS-PSP,” “variant PSP,” “PSP-parkinsonism,” “PSP-PGF,” “PSP-CBS,” “PSP-SL,” “phenotype,” and “diagnosis.” Original research articles, consensus statements, clinicopathological studies, diagnostic framework publications, validation studies, and relevant review articles published in English were considered. Particular emphasis was placed on seminal publications that introduced or substantially modified PSP classification and diagnostic approaches. Additional references were identified through manual review of bibliographies from included articles. Publications were screened for relevance to the development, validation, clinical application, strengths, and limitations of PSP diagnostic frameworks. The selected literature was synthesized narratively, with emphasis on the chronological evolution of diagnostic concepts, emerging recognition of phenotypic heterogeneity, and the transition from syndrome-based toward pathology-informed and biomarker-supported diagnostic approaches.

3. Results

Review of the available literature identified five major milestones in the evolution of PSP diagnostic frameworks (Table 1) [4,6,7,8,9]. Over the past three decades, diagnostic concepts have progressively evolved from pathology-centered definitions toward increasingly inclusive clinical classification systems that recognize the marked phenotypic heterogeneity of PSP. Early frameworks emphasized neuropathologic confirmation and the identification of classical clinical manifestations, whereas later approaches incorporated evidence from clinicopathological studies demonstrating that PSP may present with diverse motor, cognitive, behavioral, speech-language, and gait phenotypes. This shift culminated in the development of multidimensional diagnostic models that improved sensitivity for atypical and early-stage presentations while preserving diagnostic specificity. More recent refinements have focused on optimizing phenotype allocation and standardizing classification among patients who fulfill multiple diagnostic categories, reflecting the growing appreciation of PSP as a clinicopathological spectrum rather than a single, uniform clinical syndrome.

4. Historical Development of PSP Diagnostic Frameworks

4.1. Early Clinicopathological Description of PSP

PSP was first described by Steele, Richardson, and Olszewski in 1964 as a distinct neurodegenerative syndrome [1]. Based on a clinicopathological series of nine patients, the authors recognized a unique constellation of neurological features that differed from PD and other movement disorders, establishing what later became known as “Steele-Richardson-Olszewski syndrome [11].” Their seminal description, “Progressive Supranuclear Palsy: A Heterogeneous Degeneration Involving the Brain Stem, Basal Ganglia and Cerebellum With Vertical Gaze and Pseudobulbar Palsy, Nuchal Dystonia and Dementia,” emphasized supranuclear impairment of vertical gaze, particularly downgaze, a staring facial expression, pseudobulbar dysfunction, dysarthria, retrocollis with nuchal dystonia, and progressive gait impairment, while cognitive decline generally remained mild to moderate. Neuropathological examination demonstrated neuronal loss, gliosis, NFTs, granulovacuolar degeneration, and demyelination involving the BG, brainstem, and cerebellar dentate nucleus, with prominent involvement of the subthalamic nucleus, SN, superior colliculi, periaqueductal gray matter, and pontine tegmentum [1].

4.2. Pre-NINDS-SPSP Clinical Criteria for PSP

Prior to the introduction of the NINDS-SPSP criteria in 1996, several clinical frameworks were proposed to facilitate the diagnosis of PSP [10], largely based on expert clinical observations rather than systematic clinicopathological validation (Table 2)[12,13,14]. Early criteria developed by Lees emphasized a progressive nonfamilial disorder characterized by vertical SGP, particularly affecting downgaze, in combination with postural instability, pseudobulbar palsy, parkinsonism, frontal lobe dysfunction, or axial rigidity [12]. Subsequently, Golbe et al. proposed a more structured approach requiring progressive parkinsonism with SGP and additional features such as dysarthria, dysphagia, retrocollis, early falls, poor levodopa responsiveness, and minimal tremor, while incorporating exclusion criteria to distinguish PSP from other atypical parkinsonian syndromes [13]. Blin et al. later introduced separate categories of probable and possible PSP, with probable PSP requiring a highly restrictive constellation of parkinsonism, postural instability, supranuclear ophthalmoplegia, pseudobulbar symptoms, frontal dysfunction, and poor levodopa response, whereas possible PSP allowed diagnosis when most, but not all, characteristic features were present [14]. Although these frameworks established the clinical importance of vertical gaze abnormalities and early postural instability as hallmark manifestations of PSP, their diagnostic performance varied considerably, with stricter criteria providing high specificity but limited sensitivity and less restrictive criteria improving case detection at the expense of diagnostic accuracy. These limitations, together with the lack of consistent neuropathological validation, highlighted the need for a standardized evidence-based diagnostic framework and ultimately led to the development of the NINDS-SPSP criteria.

4.3. Neuropathologic Definition: Hauw Criteria (1994)

A major milestone in the evolution of PSP classification was the publication of the preliminary NINDS neuropathologic criteria by Hauw et al. in 1994, which established the first standardized pathological framework for diagnosing PSP at autopsy [9]. These criteria classified cases as typical, atypical, or combined PSP and made the semiquantitative assessment of NFTs and neuropil threads the cornerstone of diagnosis (Table 3)(Figure 1)[9]. One of the most important innovations was the recognition that tau-positive astrocytic lesions, later widely known as tufted astrocytes [15], supported the diagnosis and represented a distinctive pathological hallmark of PSP. The criteria also emphasized the characteristic distribution of tau pathology, involving the pallidum, subthalamic nucleus, striatum, nucleus basalis of Meynert, superior colliculi, periaqueductal gray matter, red nucleus, pontine tegmentum, raphe nuclei, inferior olives, and dentate nucleus, with relative sparing of the occipital cortex and cerebellar cortex. Furthermore, the framework introduced exclusion criteria to distinguish PSP from competing neurodegenerative disorders, including AD, Lewy body diseases, MSA, Pick disease, CBD, and prion disorders. The major strength of the Hauw criteria was the establishment of a reproducible neuropathological gold standard that improved diagnostic consistency among brain banks and clinicopathological studies, providing a foundation for subsequent validation of clinical criteria. However, the criteria were inherently limited by their dependence on postmortem examination, precluding their use for ante-mortem diagnosis, and by their focus on advanced pathological stages, offering little guidance for early disease detection or the recognition of the full clinical heterogeneity that would later be encompassed by variant PSP and the MDS-PSP framework.

4.4. NINDS-SPSP Criteria (1996)

The NINDS-SPSP criteria, developed by Litvan et al. in 1996, represented the first evidence-based and clinicopathologically validated clinical diagnostic framework for PSP and rapidly became the reference standard for research and clinical practice [4,10]. Using a large series of neuropathologically confirmed PSP and non-PSP neurodegenerative disorders, the authors identified vertical SGP, particularly involving downward gaze, and postural instability with unexplained falls as the clinical features most strongly associated with underlying PSP pathology. The criteria established two levels of diagnostic certainty: possible PSP and probable PSP (Figure 2)(Table 4) [10]. Probable PSP required the presence of both vertical SGP and prominent postural instability with falls occurring within the first year of symptom onset, whereas possible PSP could be diagnosed in patients with either SGP or slowed vertical saccades accompanied by early postural instability and falls. Additional supportive features included symmetrical akinetic-rigid parkinsonism, axial rigidity exceeding limb rigidity, dysarthria, dysphagia, frontal lobe dysfunction, and poor responsiveness to levodopa, while exclusion criteria were designed to eliminate competing diagnoses. Validation against neuropathologic diagnoses demonstrated excellent specificity, ranging from approximately 95% to 100% for probable PSP and from 80% to 93% for possible PSP, with positive predictive values exceeding 75% in most analyses [4,11]. However, sensitivity was substantially lower, particularly during early disease stages, with sensitivity estimates of approximately 24% to 56% at initial clinical evaluation and 39% to 83% later in the disease course [4,10,11]. These limitations reflected the criteria's strong emphasis on the classic Richardson syndrome phenotype, resulting in underrecognition of patients presenting with other variant PSP phenotypes that were subsequently recognized through clinicopathologic studies. Despite these shortcomings, the NINDS-SPSP criteria provided the first robust clinicopathological framework for ante-mortem PSP diagnosis.

5. Recognition of Clinical Heterogeneity

5.1. PSP-P

A major advance in understanding the clinical heterogeneity of PSP came from the clinicopathological study by Williams et al. in 2005, which identified PSP-P as a distinct phenotype accounting for approximately one-third of pathologically confirmed PSP cases [16]. PSP-P typically presents with asymmetric onset, bradykinesia, tremor, non-axial dystonia, and an initial moderate response to levodopa, often leading to an early clinical diagnosis of PD [17]. Patients with PSP-P generally have a longer disease duration and later age at death than those with PSP-RS, while the hallmark ocular motor abnormalities and postural instability tend to emerge later in the disease course [18]. The recognition of PSP-P demonstrated that many patients with pathologically confirmed PSP do not initially exhibit the classic PSP-RS phenotype emphasized by the NINDS-SPSP criteria, helping explain the limited sensitivity of existing diagnostic frameworks. Clinically, the identification of PSP-P broadened the recognized spectrum of PSP, highlighted the need for careful longitudinal assessment of patients with atypical parkinsonism, and provided early evidence that PSP comprises multiple phenotypic presentations rather than a single uniform clinical syndrome.

5.2. PAGF/PSP-PGF

In 2007, Williams et al. further expanded the clinical spectrum of PSP by characterizing PAGF, later incorporated into the PSP-PGF phenotype [19]. PAGF is characterized by the gradual onset of gait initiation failure and prominent FOG, often accompanied by micrographia, hypophonia, and speech freezing, while lacking the classic features of PSP during the early disease stage [20]. Diagnostic criteria included early FOG or speech, absence of limb rigidity and resting tremor, lack of a sustained response to levodopa, and the absence of dementia, supranuclear ophthalmoplegia, or significant cerebrovascular disease during the first 5 years of illness. In a clinicopathological series, six of seven patients fulfilling these criteria had pathologically confirmed PSP, yielding a positive predictive value of 86% for PSP-tau pathology [19]. Compared with PSP-RS, PAGF exhibited a substantially longer disease duration, delayed emergence of falls, gaze palsy, and cognitive impairment, and generally milder tau burden with relative sparing of the pontine base and cerebellar dentate nucleus [21].

5.3. Frontal and Cognitive Variants

Several clinicopathological studies described patients with pathologically confirmed PSP who initially presented with a syndrome resembling bvFTD, characterized by apathy, social withdrawal, loss of initiative, personality change, impaired judgment, perseverative behavior, and executive dysfunction in the absence of early SGP or postural instability [22,23]. More detailed neuropathological investigations subsequently demonstrated that these cognitive and behavioral manifestations were associated with greater involvement of frontal and allocortical regions, including the prefrontal cortex, compared with classical PSP-RS [24,25,26]. Executive dysfunction emerged as a particularly consistent feature, involving impairments in planning, cognitive flexibility, verbal fluency, attention, working memory, and problem-solving, reflecting disruption of frontal-subcortical circuits by PSP-related tau pathology [25,27].

5.4. Speech and Language Variants

Clinicopathological studies demonstrated that some patients with pathologically confirmed PSP initially present with PNFA or AOS, often several years before developing the characteristic features of SGP, postural instability, and falls [28]. Patients with PNFA typically exhibit slow, effortful, and grammatically simplified speech, impaired sentence construction, phonemic errors, and reduced verbal fluency while relatively preserving language comprehension in the early stages of disease [28]. In contrast, AOS is characterized by impaired motor planning of speech, resulting in distorted articulation, segmented speech, abnormal prosody, and difficulty initiating verbal output without a primary language deficit [28]. Subsequent clinicopathological investigations identified PSP pathology as one of the most common underlying substrates of progressive AOS and a recognized cause of PNFA, reflecting involvement of dominant frontal speech and language networks in addition to the characteristic subcortical tau pathology [28,29].

5.5. PSP-CBS

Patients with PSP-CBS typically present with asymmetric parkinsonism, limb rigidity, dystonia, ideomotor apraxia, cortical sensory deficits, myoclonus, or alien limb phenomena, often leading to an initial diagnosis of CBS rather than PSP [30]. Clinicopathological studies subsequently demonstrated that a substantial proportion of patients meeting clinical criteria for CBS harbor PSP pathology at autopsy, highlighting the poor clinicopathological specificity of the syndrome [31,32]. Compared with classical PSP-RS syndrome, PSP-CBS patients often lack early falls and SGP, delaying recognition of the underlying tauopathy and reducing the sensitivity of earlier diagnostic frameworks such as the NINDS-SPSP criteria. Pathologically, these cases exhibit characteristic PSP-type tau pathology despite a CBS presentation, reinforcing the concept that clinical syndromes and underlying neuropathology do not always correspond directly.

5.6. Cerebellar and Other Rare Variants

Patients with PSP-C typically present with gait and limb ataxia, dysarthria, impaired coordination, and postural instability as predominant early manifestations, often resulting in an initial diagnosis of MSA-C, sporadic adult-onset ataxia, or other cerebellar syndromes rather than PSP [33]. In many cases, the characteristic features of PSP, including SGP, axial rigidity, frontal dysfunction, and recurrent falls, emerge only later in the disease course, creating substantial diagnostic uncertainty during the early stages. Clinicopathological studies have demonstrated that these patients harbor typical PSP-Type 4R tau pathology despite their atypical cerebellar presentation, supporting the concept that clinical phenotype is determined by the regional distribution and severity of pathological involvement rather than by the underlying proteinopathy alone [33]. The considerable overlap between PSP-C and MSA-C is particularly important because both disorders may present with progressive gait ataxia and balance impairment; however, the relative absence of prominent autonomic failure and the later development of characteristic PSP ocular motor abnormalities may provide diagnostic clues.

6. vPSP Framework

6.1. Rationale for a New Classification System

By the early 2010s, accumulating clinicopathological evidence demonstrated that a substantial proportion of patients with pathologically confirmed PSP did not present with the PSP-RS phenotype emphasized by the NINDS-SPSP criteria. Studies identifying PSP-P, PSP-PGF, frontal behavioral syndromes, speech-language disorders, CBS, and cerebellar presentations revealed marked clinical heterogeneity and highlighted the limited sensitivity of existing diagnostic frameworks, particularly during the early stages of disease [34]. In addition, many patients failed to satisfy established phenotype definitions because they exhibited overlapping clinical features or evolved from one presentation to another over time. These observations created the need for a more inclusive classification system capable of capturing the broader clinical spectrum of PSP while maintaining its clinicopathological basis.

6.2. Respondek et al. Multicenter Clinicopathological Study

In 2014, Respondek et al. conducted the first large multicenter clinicopathological study examining 100 autopsy-confirmed PSP cases from five international brain banks [6]. The study demonstrated that classical PSP-RS accounted for only 24% of cases. Principal component analysis identified three major clinical constellations consisting of RS-like features (falls and SGP), parkinsonian features (tremor and asymmetry), and frontal-cognitive dysfunction, supporting the existence of multiple clinically meaningful PSP presentations. The authors proposed a predominance-type model based on the principal clinical features during the first two years of disease, allowing more comprehensive characterization of PSP heterogeneity and facilitating earlier recognition of atypical presentations.

6.3. Major Variant PSP Phenotypes

Building on previous clinicopathological observations, the vPSP framework incorporated several major phenotypic presentations of PSP, including PSP-RS, PSP-P, PSP-PGF, PSP-F, PSP-SL, and PSP-CBS (Table 5) [6]. Rather than viewing these syndromes as distinct diseases, the framework conceptualized them as different clinical manifestations of a common underlying PSP tauopathy. Importantly, the model recognized that individual patients may display overlapping features and undergo phenotypic evolution during the disease course, emphasizing a spectrum-based rather than categorical approach to classification. The framework also demonstrated that survival, cognitive impairment, and diagnostic latency differed significantly among phenotypes, highlighting their prognostic relevance.

7. MDS-PSP Criteria (2017)

7.1. Development of the MDS-PSP Criteria

The MDS-PSP criteria were developed in response to the recognized limitations of the NINDS-SPSP framework, which demonstrated excellent specificity for PSP-RS but insufficient sensitivity for early disease and non-Richardson phenotypes. To address these shortcomings, the MDS–endorsed PSP Study Group conducted a comprehensive evidence-based and consensus-driven revision process that incorporated systematic literature review, analysis of more than 200 autopsy-confirmed PSP cases and multiple disease controls, expert panel discussions, and repeated Delphi consensus rounds. The primary objective was to improve recognition of early and vPSP presentations while preserving diagnostic specificity and maintaining a close relationship with the underlying neuropathological definition of PSP as a 4R tauopathy (Figure 3)(Table 6)[7].

7.2. Conceptual Innovations

A major innovation of the MDS-PSP criteria was the transition from a syndrome-based model focused primarily on PSP-RS to a multidimensional framework capable of capturing the broader phenotypic spectrum of PSP. Instead of relying predominantly on early falls and SGP, the new criteria recognized four core functional domains representing the principal manifestations of PSP pathology: ocular motor dysfunction (O), postural instability (P), akinesia (A), and cognitive dysfunction (C) (Table 7) [7]. Within each domain, clinical features were stratified according to their diagnostic value, allowing flexible combinations of findings to generate syndrome-specific diagnoses and varying levels of diagnostic certainty. This approach acknowledged that PSP may initially present through motor, gait, cognitive, behavioral, speech-language, or cortical syndromes rather than a single stereotyped phenotype.

7.2.1. Ocular Motor Domain (O)

The ocular motor domain identifies characteristic abnormalities of vertical eye movement control and includes vertical SGP, slowed vertical saccades, and frequent macro square-wave jerks or eyelid opening apraxia. These features reflect the prominent involvement of midbrain gaze centers and remain among the most specific clinical indicators of underlying PSP pathology.

7.2.2. Postural Instability Domain (P)

The postural instability domain emphasizes early impairment of balance and locomotion, including recurrent unprovoked falls, a tendency to fall during the pull test, and abnormal postural correction responses. Incorporation of graded postural features allowed earlier recognition of patients presenting with predominant gait and balance disturbances before the emergence of classical PSP-RS.

7.2.3. Akinesia Domain (A)

The akinesia domain captures the heterogeneous parkinsonian manifestations of PSP, ranging from progressive gait freezing and axial levodopa-resistant akinetic-rigid syndrome to asymmetric, tremor-predominant, or partially levodopa-responsive parkinsonism. This domain facilitated recognition of PSP-P and PSP-PGF phenotypes that were frequently underdiagnosed by previous criteria.

7.2.4. Cognitive Dysfunction Domain (C)

The cognitive dysfunction domain incorporates frontal behavioral syndromes, speech-language disorders, and corticobasal syndrome presentations. Included features comprise executive dysfunction, apathy, disinhibition, progressive nonfluent aphasia, apraxia of speech, and corticobasal cortical signs. The addition of this domain formally acknowledged the cortical manifestations of PSP and expanded diagnostic applicability beyond traditional movement disorder presentations.

7.3. Levels of Diagnostic Certainty

To accommodate varying clinical presentations and disease stages, the MDS criteria introduced a hierarchical system of diagnostic certainty (Table 8) [7]. Definite PSP remains restricted to neuropathological confirmation. Probable PSP identifies patients with highly specific clinical combinations suitable for clinical trials and biomarker studies. Possible PSP increases sensitivity by capturing broader phenotypic presentations, whereas Suggestive of PSP was introduced as a novel category intended to facilitate recognition of early or incomplete syndromes before full diagnostic criteria are satisfied. This structure represented a major shift toward earlier diagnosis and prospective identification of patients suitable for disease-modifying interventions. Another important innovation was the introduction of the category probable 4R-tauopathy, which includes patients fulfilling criteria for PSP-SL or PSP-CBS. This category acknowledges the strong association of these clinical syndromes with underlying 4R tau pathology while recognizing the difficulty of differentiating PSP from corticobasal degeneration during life.

7.4. Phenotype-Specific Diagnostic Categories

Unlike earlier frameworks, the MDS criteria explicitly incorporated multiple PSP phenotypes under a unified diagnostic system. These included PSP-RS, PSP-P, PSP-PGF, PSP-OM, PSP-PI, PSP-CBS, PSP-F, and PSP-SL. Importantly, the criteria viewed these syndromes as alternative clinical expressions of a common underlying PSP pathology rather than as distinct diseases. This phenotype-based approach reflected the growing recognition that most patients with autopsy-confirmed PSP do not initially present with classical PSP-RS.

8. Phenotype Allocation and MAX Rules

8.1. Challenges of Multiple Phenotype Assignment

Although the 2017 MDS-PSP criteria represented a major advance in recognizing the clinical heterogeneity of PSP, their broad phenotype-inclusive structure introduced a practical challenge: individual patients frequently fulfilled criteria for multiple PSP phenotypes simultaneously. Because the criteria incorporate four clinical domains (O-P-A-C), combinations of findings can generate several overlapping phenotype assignments in a single patient, particularly later in the disease course as additional manifestations emerge. In a neuropathologically confirmed cohort, more than 90% of PSP patients fulfilled criteria for multiple phenotypes, with an average of approximately seven phenotype allocations per patient [8]. This multiplicity complicated clinical classification, hindered standardization across studies, and created challenges for patient stratification in observational studies and therapeutic trials. Furthermore, phenotypic evolution over time meant that patients initially presenting with one syndrome could subsequently develop features characteristic of several others, making consistent diagnostic labeling increasingly difficult.

8.2. Development of Maximum Allocation Extinction Rules

To address these limitations, the MDS-endorsed PSP Study Group developed the MAX rules, a set of hierarchical principles designed to assign a single predominant phenotype to each patient despite the presence of multiple qualifying diagnoses. The MAX framework prioritizes phenotypes according to diagnostic certainty, temporal evolution, and their predictive value for underlying PSP pathology (Figure 4)[8]. First, the phenotype associated with the highest level of diagnostic certainty is retained. Second, the earliest clinically predominant syndrome is favored until another phenotype clearly becomes dominant. Third, phenotypes are ranked hierarchically according to their association with PSP neuropathology, with PSP-RS generally receiving highest priority, followed by PSP-OM and PSP-PI, and subsequently other variant phenotypes. Application of these rules substantially reduced diagnostic multiplicity, yielding a single predominant phenotype in the majority of patients while preserving clinically relevant information about disease evolution.

9. Comparative Analysis

9.1. Evolution of Diagnostic Concepts

The evolution of PSP diagnostic frameworks reflects a progressive shift from pathology-based definitions toward multidimensional clinical classification systems (Table 9) [4,6,7,8,9]. The Hauw criteria established a neuropathological gold standard for PSP diagnosis, whereas the NINDS-SPSP criteria introduced the first clinicopathologically validated ante-mortem diagnostic approach centered on the classical PSP-RS phenotype. Subsequent recognition of multiple non-Richardson presentations led to the development of the vPSP framework, which conceptualized PSP as a spectrum of phenotypes rather than a single clinical syndrome. These observations ultimately informed the MDS-PSP criteria, which integrated phenotypic diversity into a unified diagnostic system through the use of core functional domains and graded levels of diagnostic certainty.

A central challenge in PSP diagnosis has been balancing diagnostic sensitivity and specificity. The NINDS-SPSP criteria demonstrated excellent specificity, particularly for probable PSP, but limited sensitivity during the early disease stages and among variant phenotypes. Recognition of PSP-P, PSP-PGF, PSP-F, PSP-SL, and PSP-CBS highlighted the inability of earlier criteria to identify a substantial proportion of pathologically confirmed cases. By incorporating multiple phenotypic presentations and introducing the "suggestive of PSP" category, the MDS-PSP criteria substantially improved sensitivity while maintaining acceptable specificity. Consequently, modern frameworks provide a better balance between accurate diagnosis and early case detection.

9.2. Remaining Challenges and Future Needs

Despite substantial advances, important challenges remain (Table 10). Definite diagnosis still requires neuropathological confirmation, and considerable overlap persists with CBD, MSA, FTD, and other atypical parkinsonian disorders. Furthermore, phenotype evolution over time may complicate classification even within the MDS-PSP framework, necessitating approaches such as the MAX rules. Future diagnostic systems will likely incorporate imaging, fluid, genetic, and molecular biomarkers to improve early diagnosis, enhance pathological specificity, and facilitate precision medicine approaches in PSP.

10. Emerging Biomarkers and Future Directions

10.1. MRI Biomarkers

Structural MRI has become one of the most extensively investigated imaging modalities for PSP and serves as an important adjunct to clinical diagnosis. Characteristic findings include selective midbrain atrophy resulting in the so-called hummingbird sign on sagittal imaging and the morning glory sign on axial images. Although highly specific in advanced disease, these visual signs have limited sensitivity during early stages and are therefore insufficient as standalone diagnostic markers [35]. The MRPI, calculated from measurements of the midbrain, pons, middle cerebellar peduncle, and superior cerebellar peduncle, reliably distinguishes PSP from PD and MSA [36]. Subsequent refinement led to the development of MRPI 2.0 [37], which incorporates third ventricle width and frontal horn measurements and has shown improved performance for identifying PSP-P and other non-RS phenotypes.

10.2. Fluid Biomarkers

NfL is currently the most consistently validated fluid biomarker in PSP. Elevated CFS and plasma NfL concentrations reflect axonal injury and correlate with disease severity, progression rate, and survival. Although NfL lacks specificity for PSP, it may assist in distinguishing atypical parkinsonian syndromes from PD and serve as a valuable prognostic marker in clinical trials [38]. Because PSP is a primary 4R tauopathy, considerable effort has focused on developing tau-based biomarkers. However, conventional CSF total tau and phosphorylated tau measurements have generally shown limited diagnostic value, partly because they were originally developed for AD [39]. More recently, assays targeting tau fragments, conformations, and 4R tau-specific species have shown promise, although none have yet reached routine clinical implementation [40]. The development of PSP-specific tau biomarkers remains a critical unmet need.

10.3. Tau PET Imaging

Tau PET imaging represents one of the most promising approaches for in vivo identification of PSP pathology. First-generation tracers such as 18F-AV-1451 (flortaucipir) demonstrated increased uptake in the globus pallidus, subthalamic nucleus, and midbrain of PSP patients but were limited by substantial off-target binding and reduced affinity for non-AD tau aggregates [41]. Second-generation tau PET ligands have shown improved binding characteristics and may better differentiate PSP from other neurodegenerative disorders [42]. Nevertheless, challenges remain regarding sensitivity, specificity, and correlation with neuropathological tau burden. At present, tau PET remains primarily a research tool but has the potential to transform diagnosis by enabling direct visualization of PSP-related tau pathology during life.

10.4. Implications for Clinical Trials

The emergence of biomarker-supported diagnosis has important implications for therapeutic development. Historically, reliance on clinical criteria alone resulted in delayed diagnosis and enrollment of heterogeneous patient populations, potentially reducing the ability of trials to detect treatment effects. Earlier and more accurate identification of PSP through biomarkers could facilitate recruitment during stages when neurodegeneration remains potentially modifiable. Biomarkers may also enable patient stratification according to phenotype, pathological burden, and disease progression rate, thereby supporting precision medicine approaches. Furthermore, quantitative imaging and fluid biomarkers could serve as surrogate outcome measures, providing objective assessments of target engagement and disease modification. As disease-modifying therapies targeting tau pathology advance into clinical testing, the integration of biomarkers into diagnostic and trial frameworks will likely become essential for optimizing study design and therapeutic success.

11. Conclusions

The evolution of PSP diagnostic frameworks reflects a broader transformation in the understanding of PSP from a relatively uniform clinicopathological entity into a heterogeneous spectrum of clinical phenotypes linked by a common underlying 4R tauopathy. The Hauw criteria established a neuropathological gold standard for diagnosis, whereas the NINDS-SPSP criteria provided the first clinicopathologically validated ante-mortem framework with excellent specificity for classical PSP-RS. Subsequent clinicopathological studies demonstrated that many patients with pathologically confirmed PSP present with alternative motor, cognitive, behavioral, speech-language, and cortical syndromes, leading to the development of the vPSP framework and fundamentally reshaping concepts of disease classification. These observations directly informed the 2017 MDS-PSP criteria, which incorporated multiple phenotypes, four core functional domains, supportive diagnostic features, and graded levels of diagnostic certainty, substantially improving recognition of atypical and early-stage disease. The later introduction of the MAX rules further standardized phenotype allocation in patients fulfilling multiple diagnostic categories, facilitating consistency across clinical practice and research. Despite these advances, important limitations remain, including persistent clinicopathological overlap with related neurodegenerative disorders, phenotype evolution over time, and the continued reliance on clinical manifestations that often emerge after significant neurodegeneration has occurred. Emerging MRI markers, fluid biomarkers, and tau-targeted molecular imaging hold considerable promise for improving early diagnosis, pathological specificity, patient stratification, and therapeutic trial design. Consequently, future diagnostic paradigms will likely integrate clinical phenotyping with multimodal biomarkers, moving toward biologically informed and pathology-based classification systems capable of supporting precision medicine approaches and disease-modifying therapies in PSP.