Clinical research criteria for the diagnosis of progressive supranuclear palsy (Steele-Richardson-Olszewski syndrome)

Progressive supranuclear palsy (PSP), also called Steele-Richardson-Olszewski syndrome, is one of the most common atypical parkinsonian syndromes. PSP was well described as a clinicopathologic entity in 1964, ^[1,2] although a few single clinical or pathologic reports were previously published. ^[3-6] The estimated prevalence of PSP is 1.39 per 100,000 inhabitants after age adjustment to the U.S. population. ^[7,8] However, this is probably a considerable underestimate, as diagnosed cases are often not recognized until the disease has run half its course, ^[9] and probably many more patients die without a diagnosis, or have a misdiagnosis, usually Parkinson's disease (PD).

Currently, there is no biologic marker for the diagnosis of PSP. Neuropathologic examination remains the ``gold standard'' for its definitive diagnosis. Preliminary National Institute of Neurological Disorders and Stroke (NINDS) neuropathologic criteria for the diagnosis of PSP and related disorders were recently proposed. ^[10,11] The NINDS neuropathologic criteria ^[11] specified two subtypes of PSP: typical and combined. (The subtype of atypical PSP originally proposed ^[10] is omitted from this discussion because it was found not to be reliable. ^[11]) Typical PSP conforms to the original description of Steele et al. ^[1] The neuropathologic characteristics of PSP include a high density of neurofibrillary tangles and neuropil threads in the basal ganglia and brainstem with a characteristic distribution. ^[10-14] Neuropil threads are filamentous structures scattered throughout the neuropil, occurring independently of neurofibrillary tangles and indicating a diffusely distributed disease of the cytoskeleton. ^[12] Tau-positive astrocytes or processes in areas of involvement help to confirm the diagnosis. ^[10,13,14] These characteristic pathologic findings are accompanied by a relatively nonspecific and variable neuronal loss and gliosis. Combined PSP is restricted to the coexistence of neuropathologically typical PSP and vascular lesions in the brainstem and basal ganglia. ^[11] Otherwise, when vascular lesions are found in areas distinct from these characteristically affected in PSP, or other CNS disorders occur, the diagnosis should be typical PSP plus the associated disease. ^[11,15] Patients with a neuropathologic diagnosis of PSP may also have other associated neurodegenerative disorders. ^[16] Neuropathologists, therefore, may be routinely confronted with complicated PSP cases, and could benefit from the use of standardized criteria for PSP diagnosis.

Neuropathologically confirmed cases of corticobasal ganglionic degeneration (CBGD), dementia with Lewy bodies, cerebrovascular disease, subcortical gliosis, prion disease, and multiple system atrophy have been clinically misdiagnosed as PSP. ^[17-25] In a recent study, ^[26] in which retrospectively collected and abstracted clinical and laboratory data on 105 cases of parkinsonism or dementia were presented to six neurologists for ``blind'' diagnosis, false-positive diagnoses accounted for up to 24 to 50% of the PSP diagnoses. In addition, neuropathologically confirmed cases of PSP may be confused clinically with other syndromes in which the clinical features overlap with those of PSP, such as PD, CBGD, multiple system atrophy, and Alzheimer's disease (AD). ^[27-31] Misdiagnosis not only hinders the study of PSP, but also work on disorders with which it is confused. Currently, there is no effective treatment for PSP. ^[32]

The recruitment of a population of patients with an accurate diagnosis of PSP is essential for therapeutic and epidemiologic research and for the confident comparison of the findings from different studies. This effort depends on the availability of standard clinical criteria for the diagnosis of PSP. There are several published sets of criteria for the clinical diagnosis of PSP ^[33-38] but, though some of them are reliable, they incorporate many falsenegative or false-positive cases. ^[26] On May 5 and 6, 1995, the NINDS and the Society for Progressive Supranuclear Palsy, Inc. (SPSP) sponsored the NINDS-SPSP international workshop to develop an accurate and universally accepted set of criteria for the diagnosis of PSP.

This paper presents the consensus findings on the diagnosis of PSP resulting from the workshop and reviews the main studies that led to their development. We describe the common and uncommon clinical features of PSP, review the validity of previously published criteria, and report on the validity of the NINDS-SPSP clinical criteria as determined in a well-characterized data set.

Common clinical features of PSP.

The medical history and physical examination are most relevant for making the clinical diagnosis of PSP, but the lack of systematic evaluation of patients and overlap of symptoms make the early diagnosis difficult. Therefore, the following description of the typical clinical features of PSP is based mainly on cases with known neuropathology. ^[39-42]

The onset of symptoms in PSP is insidious, and the evolution of symptoms may vary. The mean age at PSP onset ranges from 55 to 70 years, ^[39-42] with a few cases beginning as early as 45 years of age. The first symptoms of PSP are usually postural instability and falls, which occur either at the onset of the disease (in 63% of the cases reviewed ^[42]) or during the first year (in 69% of the cases ^[39]). Cognitive or behavioral changes usually begin in the first year (in 52% of the cases ^[39]), but rarely occur at the disease onset (in 8% of the cases ^[42]). After postural instability, dysarthria is the second most common symptom of PSP (in 33% of cases at onset ^[42] and 40% during the first year ^[39]), and bradykinesia is the third most common problem (in 13% of cases at onset ^[42] and 22% during the first year ^[39]).

Visual disturbances (e.g., diplopia, blurred vision, burning eyes, and light sensitivity) are also early symptoms (in 13% of newly diagnosed PSP cases ^[42] and during the first year in 64% ^[39]). Characteristically, supranuclear gaze deficits in PSP involve initially either downward or upward gaze and, later, horizontal gaze. Vertical supranuclear palsy, the hallmark of the disease, may be present at the disease onset, but in many cases may only be detected several years later, or may never develop. ^[6,41] Because limitation of upward gaze is more common than limitation of downward gaze in neurodegenerative disorders, and because there is also a certain degree of upward gaze restriction with aging, limitation of downward gaze is considered more specific for the diagnosis of PSP. ^[33,41,43]

Early in the course of the disease, pursuit eye movements may become saccadic; voluntary saccades may become slow with preserved range of motion, or hypometric saccades may develop. In addition, patients may fail to suppress the vestibuloocular reflex, and may have abnormalities in convergence. ^[44] Later signs include slowing of eye opening or eye closure, slowing of eyelid saccades (i.e., eyelid movement occurring when subjects with supranuclear palsy attempt to make a vertical saccade), and an inability to suppress a blink to a bright light, (i,e., a ``visual'' rather than a somatosensory glabella, or Meyerson, sign) (personal communication, D.S.Z.).

The first clinical visit to a tertiary center occurs a mean of 1.5 to 3.7 years after the onset of PSP symptoms. ^[42] In 16 cases of PSP confirmed by pathologic examination, Colosimo et al. ^[41] reported that within 3 years of onset 94% had bilateral parkinsonism and instability, 81% symmetrical onset, 50% vertical supranuclear palsy, 56% axial ``dystonia,'' and 50% had cognitive impairment. At the first visit, in the study by Litvan et al., <sup>[42]</sup> PSP patients had gait disorder and postural instability, a history of falls, bilateral bradykinesia, a predominant akinetic-rigid disease course, vertical supranuclear palsy, dysarthria, dysphagia, axial rigidity, neck dystonia, frontal lobe type symptoms, and personality changes involving mostly apathy and depression Table 1. The presence of ``staring,'' nonblinking facies (in 92% of cases), and sitting ``en bloc'' (in 30% of cases) are also characteristic of PSP. <sup>[38]</sup> In contrast with the hypomimia of PD, the facial expression of patients with PSP is often described as ``worried'' or ``astonished.'' ^[45]

In general, the early symptoms and signs of PSP steadily worsen. Gait abnormalities with postural instability and bradykinesia are observed in most patients at the last visit Table 1. The mean survival of PSP from disease onset is 5 to 6.7 years (range, 1 to 13 years) in autopsy series ^[38-42] and 5.9 to 6.9 years (range 1 to 17) in clinical ones. ^[7-9] The most common cause of death is pneumonia. ^[42]

Uncommon clinical features of PSP.

Features of PSP that deviate from the usual disease course, and do not support the diagnosis, include onset of symptoms before age 40 or a disease duration of more than 20 years. However, there are other uncommon features that do not exclude the diagnosis of PSP: late onset of postural instability and falls, mild asymmetric parkinsonism, and absence of cognitive disturbances. ^[38,42] Infrequently, PSP may present with absence of gaze restriction, pure akinesia, severe dementia, or a family history of PSP. ^[46-55] It is unclear whether these unusual features have any biologic significance since, in a small series, ^[56] no histologic distinction could be made between cases of PSP with vertical supranuclear palsy, a sign that is considered to be essential for the diagnosis, and cases without it.

Proposed NINDS-SPSP criteria.

The NINDS-SPSP criteria were developed in two phases. In the first phase, previously published criteria for the clinical diagnosis of PSP ^[33-35] were validated by six neurologists, who were unaware of the diagnosis, using clinical information retrospectively abstracted from clinical records of patients with autopsy-confirmed cases of dementia or parkinsonism provided by seven medical centers specializing in neurodegenerative disorders. ^[26] The diagnoses of these cases were based on the NINDS neuropathologic criteria, ^[11] which have been shown to have substantial interobserver reliability. Because the previously published clinical criteria ^[33-35] were found to be suboptimal, new preliminary clinical criteria were proposed and validated in the same data set. ^[26]

In the second phase, the preliminary clinical criteria were revised by the consensus of experts (movement disorder specialists, neuroophthalmologists, behavioral neurologists, and neuropsychologists) experienced with the diagnosis of PSP. These experts, meeting in plenary session, refined the criteria by characterizing the common and uncommon clinical features of PSP, as discussed earlier, differentiating the features of PSP and related disorders Table 2, and reviewing the relevant literature. Finally, NINDS-SPSP criteria, specifying mandatory inclusion and exclusion criteria, as well as supportive criteria, were adopted Table 3. These criteria were evaluated in a well-characterized data set containing information commonly available in carefully made records. The data set comprised 83 cases of autopsy-confirmed parkinsonian disorders Table 4, including PSP (n = 24), CBGD (n = 11), PD (n = 11), dementia with Lewy bodies (n = 14), multiple system atrophy (n = 15), and Pick's disease (n = 8). Patients were selected from the research and clinical neuropathologic files of seven medical centers in Europe and the United States by neuropathologists ^[42] who used the NINDS neuropathologic criteria for the diagnosis of PSP and related disorders for their diagnoses. ^[10,11] The patient records were abstracted by neurologists who followed strict instructions such as to record as missing any features that were not explicitly described in the records and to record clinical descriptions uniformly, according to specific definitions provided. The sensitivity, specificity, and positive predictive values for the diagnosis of PSP were chosen as measures of validity. Sensitivity is defined as the percentage of cases with a clinical diagnosis of PSP using a set of criteria among all PSP cases confirmed by neuropathologic examination. Specificity is the percentage of cases without a clinical diagnosis of PSP among all non-PSP cases confirmed by neuropathologic examination. Positive predictive value is the percentage of autopsy-confirmed PSP cases among all cases with a clinical diagnosis of PSP using a set of criteria. The validity of the NINDS-SPSP criteria, compared with that of previously proposed criteria, ^[33-38] is shown in Table 5.

Discussion.

This is the first attempt to develop criteria for the diagnosis of PSP following a programmatic effort that included the standardization of neuropathologic ^[11] as well as clinical criteria. ^[26] In spite of the thorough process used, the validity of the NINDS-SPSP criteria needs to be tested in prospective studies.

The possibility of detecting and validating the clinical criteria in a well-characterized data set with known neuropathologic diagnosis allowed us to concentrate on the early symptoms of the disease. Early postural instability with falls and neuroophthalmologic findings helped us to avoid the usual misdiagnosis of PSP as PD. Moreover, falls occurring during the first year in a patient with either upward or downward gaze supranuclear palsy, but without any of the exclusionary features, strongly indicated the diagnosis of PSP, because none of the patients with this cluster of features had any other disease. However, in our study, upward-gaze abnormalities were moderately to severely restricted (at least 50% of the normal range) to eliminate age-related impairments. Further, when we included only downward-gaze palsy as one of the criteria for probable PSP, there was no effect on specificity, but the sensitivity decreased to 38%, compared with 50% when vertical gaze palsy was used. The result was similar for possible PSP.

Even though the features proposed for the diagnosis of probable PSP are highly specific (100%), they are not sensitive enough, as they can detect only half of the patients at the first visit. Thus, the NINDS-SPSP criteria for the diagnosis of probable PSP are suitable for clinical drug trials or studies on the biology of the disease, when it is essential to exclude subjects with illnesses other than PSP. On the other hand, the NINDS-SPSP criteria for the clinical diagnosis of possible PSP, which are more sensitive (detected 83% of the patients at the first visit 3 years after onset) but less specific (false-positive ratio of 17%), are more suitable for descriptive epidemiologic studies or clinical care, when it is important to identify all, or nearly all, cases of PSP even at the cost of including a few false-positive cases.

Slowing of vertical saccades is suggested as an early sign of PSP (see Appendix), but missing data precluded us from evaluating its predictive value in this data set. However, 13% percent of the PSP patients whose condition was not detected with the NINDS-SPSP criteria for possible PSP had postural instability with falls within 1 year of onset. Therefore, alternative, untested criteria for possible PSP, in the absence of supranuclear limitation of vertical gaze, are slow vertical saccades together with early postural instability with falls Table 3. Because slowing of vertical saccades was not part of the clinical criteria tested in the present study, this feature needs to be subjected to a similar validation study.

The exclusionary criteria Table 3 are important because they provide the specificity for the diagnosis of PSP. For example, in possible PSP, the omission of the exclusionary criteria decreases the specificity to 89% and the positive predictive value to 77%.

The supportive criteria Table 3 are included to remind physicians of important features often associated with PSP, but that are neither essential nor sufficient for the diagnosis. Some of these features, such as retrocollis, occur infrequently. Others, such as a poor response to levodopa or frontal lobe symptoms, may become more relevant for the diagnosis when tested prospectively. Similarly, there may be other features, such as early palilalia, that could be identified and tested for diagnostic relevance.

The proposed NINDS-SPSP criteria improve upon previously published sets of criteria. ^[26] Prospective studies are needed to determine whether ancillary studies can further improve the diagnosis of PSP. These investigations should validate the usefulness of (1) neuropsychological studies of frontal lobe dysfunction, ^[58-63] (2) PET studies to determine frontal cortex metabolism and striatal dopamine D2-receptor function, ^[64-72] (3) recording of event-related potentials to assess normal early N1 component and delayed P2 and P300 components, ^[73-75] (4) electrophysiologic testing of the auditory startle response, ^[76,77] (5) recording of eye movements to detect early abnormalities, ^[43,44,78] (6) MRI to detect brainstem atrophy and to rule out other disorders, ^[79-81] and (7) polysomnographic studies to detect abnormalities of rapid-eye movement sleep. ^[82-85]

Ideally, the proposed NINDS-SPSP criteria will assist in identifying biologic markers that in turn will yield an earlier diagnosis of this disorder.

Acknowledgments

We wish to thank Drs. J.-J. Hauw, S. Daniel, D. Dickson, D.S. Horoupian, K. Jellinger, P.L. Lantos, and A. McKee for providing the cases for the data base, Drs. L. D'Olhaberriague, C.A. Mangone, R.K.B. Pearce, and K. Ray Chaudhuri for assisting with the study, and B.J. Hessie for skillful editing.

Appendix: Guidelines for testing.

Workshop Participants and Affiliations.

Yves Agid, MD, PhD, Workshop Co-chair, Laboratoire de Medicine Experimentale, INSERM UNIT 289, Hopital de la Salpetriere, Paris; Donald Calne, MD, Division of Neurology, Neurodegenerative Disorders Centre, University Hospital UBC Site, Vancouver; Gregory Campbell, PhD, Analytical Biometrics Section, Biometry and Field Studies Branch, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Thomas N. Chase, MD, Experimental Therapeutics Branch, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Bruno Dubois, MD, Laboratoire de Medicine Experimentale, INSERM UNIT 289, Hopital de la Salpetriere, Paris; Roger C. Duvoisin, MD, Department of Neurology, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ; Edmond J. FitzGibbon, MD, National Eye Institute, Bethesda, MD; Christopher G. Goetz, MD, Department of Neurology, Rush Medical College, Chicago, IL; Lawrence I. Golbe, MD, Department of Neurology, UMDNJ-Robert Wood Johnson Medical School, New Brunswick, NJ; Jordan Grafman, PhD, Cognitive Neuroscience Section, Medical Neurology Branch, National Institute of Neurological Disorders and Stroke, Bethesda, MD; John H. Growdon, MD, Massachusetts Alzheimer's Disease Research Center, Harvard Medical School, Massachusetts General Hospital, Boston, MA; Zach Hall, PhD, Office of the Director, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Mark Hallett, MD, Office of the Clinical Director, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Joseph Jankovic, MD, Parkinson Disease Center and Movement Disorder Clinic, Department of Neurology, Baylor College of Medicine, Houston, TX; Jorge Juncos, MD, Parkinson's and Alzheimer's Disease Center, Department of Neurology, Emory University, Atlanta, GA; Irene Litvan, MD, Workshop Co-chair, Neuroepidemiology Branch, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Karin B. Nelson, MD, Neuroepidemiology Branch, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Eugene Oliver, PhD, Division of Demyelinating, Atrophic, and Dementing Disorders, National Institute of Neurological Disorders and Stroke, Bethesda, MD; Peter Pramstaller, MD, Department of Clinical Neurology, Institute of Neurology, London; Niall P. Quinn, MD, Department of Clinical Neurology, Institute of Neurology, London; Stephen G. Reich, MD, Parkinson Disease Center, Department of Neurology, Johns Hopkins Hospital, Baltimore, MD; Eduardo Tolosa, MD, Department of Neurology, University of Barcelona Hospital Clinic, Barcelona; Marc Verny, MD, Laboratoire de Neuropathologie Raymond Escourolle, INSERM U 360, Hopital de la Salpetriere, Paris; David S. Zee, MD, Ocular Motor-Vestibular Testing Laboratory, Johns Hopkins Hospital, Baltimore, MD.