Accuracy of the National Institute for Neurological Disorders andStroke/Society for Progressive Supranuclear Palsy and Neuroprotection

and Natural History in Parkinson Plus Syndromes Criteria for theDiagnosis of Progressive Supranuclear Palsy

Gesine Respondek, MD,1,2,3 Sigrun Roeber, MD,4 Hans Kretzschmar, MD,4 Claire Troakes, PhD,5

Safa Al-Sarraj, FRCPath,5 Ellen Gelpi, MD,6 Carles Gaig, MD,6 Wang Zheng Chiu, MD,7 John C.van Swieten, MD,7

Wolfgang H. Oertel, MD,1 G€unter U. H€oglinger, MD1,2,3*

1Department of Neurology, Philipps Universit€at, Marburg, Germany2Department of Neurology, Technische Universit€at M€unchen, Munich, Germany

3German Center for Neurodegenerative Diseases, Munich, Germany4Center for Neuropathology and Prion Research, Ludwig Maximilians University, Munich, Germany

5Medical Research Council London Neurodegenerative Diseases Brain Bank, King’s College, London, UK6Neurological Tissue Bank and Neurology Department, Hospital Clinic/Institut d’Investigacio Biomedica August Pi i Sunyer, Barcelona, Spain

7Department of Neurology, Erasmus Medical Center, Rotterdam, The Netherlands

ABSTRACT: Autopsy is the diagnostic goldstandard for progressive supranuclear palsy (PSP). TheNational Institute of Neurological Disorders and Strokeand Society for Progressive Supranuclear Palsy(NINDS-SPSP) criteria for the clinical diagnosis of“probable” PSP are thought to possess high specificityand low sensitivity. The NINDS-SPSP criteria for“possible” PSP are considered to increase sensitivity atthe expense of specificity. The Neuroprotection andNatural History in Parkinson Plus Syndromes (NNIPPS)criteria are intended to improve sensitivity while main-taining high specificity. The aim of this study was toconduct a clinicopathological evaluation of the NINDS-SPSP and NNIPPS criteria in tertiary neurological cen-ters. Defined clinical features and their year of onsetwere recorded by chart review in neuropathologicallydiagnosed patients with PSP, Parkinsons’s disease (PD),MSA parkinsonism and corticobasal degeneration fromfour European brain banks. Fulfilment of the clinical

diagnostic criteria was verified for each year after dis-ease onset and for the final antemortem record. We an-alyzed 98 PSP patients and 46 disease controls. TheNINDS-SPSP “probable” criteria yielded shorter time todiagnosis, slightly higher specificity and positive predic-tive value (PPV), and similar sensitivity, compared withthe NNIPPS criteria. Unexpectedly, the NINDS-SPSP“possible” criteria yielded the lowest sensitivity, speci-ficity, and PPV. A combination of NINDS-SPSP possibleand probable criteria yielded the highest sensitivity. Wesuggest that the NINDS-SPSP probable criteria mightbe preferred for recruitment of patients for clinical trials,where an early and specific diagnosis is important. Forroutine clinical care, where high sensitivity is crucial, acombination of NINDS possible and probable criteriamight be preferred. VC 2013 Movement Disorder Society

Key Words: progressive supranuclear palsy; clinicaldiagnostic criteria; neuropathology; sensitivity; specificity

Progressive supranuclear palsy (PSP) is an adult-onsetneurodegenerative disorder with tau pathology at manylevels of the central nervous system, leading to an aki-netic-rigid syndrome with ocular motor dysfunction,

postural instability, and frontal lobe and bulbar dys-function.1 Diagnostic gold standard for “definite PSP”is the postmortem neuropathological diagnosis.2 In1996, the National Institute of Neurological Disorders

------------------------------------------------------------------------------------------------------------------------------*Correspondence to: Dr. G€unter U. H€oglinger, Department of Translational Neurodegeneration, German Center for Neurodegenerative Diseases, M€unchen,Max Lebsche Platz 30, D-81677 Munich, Germany; guenter.hoeglinger@dzne.de

Relevant conflicts of interest/financial disclosures: G.U.H. is supported by the Deutsche Forschungsgemeinschaft (HO2402/6-1). E.G. is partiallyfunded by the Spanish Ministerio de Economia y Competitividad, Programa de T�ecnicos de Apoyo 2011, and received a research grant from theAcademia de Ciencies Mediques de Catalunya 2011.

Full financial disclosures and author roles may be found in the online version of this artcle.

Received: 3 July 2012; Revised: 31 October 2012; Accepted: 26 November 2012Published online 21 February 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/mds.25327

R E S E A R C H A R T I C L E

504 Movement Disorders, Vol. 28, No. 4, 2013

and Stroke and the Society for PSP (NINDS-SPSP) crite-ria have been proposed for the clinical diagnosis ofPSP, based on the analysis of clinical features of a casemix of patients with pathologically confirmed forms ofparkinsonism.1 Briefly, the NINDS-SPSP criteria rely onthe identification of a progressive disorder with onsetafter age 40 combining the two cardinal features (i.e.,postural instability with falls during the first year of thedisease and slow vertical saccades or supranuclear gazepalsy). A list of exclusion criteria aims to sort out look-alike disorders. Two sets of the NINDS-SPSP criteriahave been proposed to achieve different levels of diag-nostic certainty: (1) “clinically probable PSP” criteriaaimed at high specificity, accepting compromised sensi-tivity, to identify patients for research purposes and (2)“clinically possible PSP” criteria aimed at high sensitiv-ity, accepting compromised specificity, to identifypatients for medical care.1 Validation of these criteriain an independent set of patients demonstrated a highpositive predictive value (PPV) for both NINDS-SPSP“possible” and “probable” criteria, albeit low sensitiv-ity, particularly during the early course of the disease.3

The time of clinical diagnosis of PSP is usually delayedby several years after disease onset, mainly because theocular motor cardinal features of the disease may de-velop late or never during the course of the disease.4

Recently, the Neuroprotection and Natural History inParkinson Plus Syndromes (NNIPPS) consortium hasproposed modified clinical diagnostic criteria, aiming tofacilitate the clinical diagnosis of PSP and to improvesensitivity while maintaining a high specificity.5 Briefly,the NNIPPS criteria allow an age of onset of �30 andpostural instability or falls within 3 years from diseaseonset. Until now, the validity of the NNIPPS criteriahas not been verified in an independent study. Thus,we aimed to compare the diagnostic accuracy of theNINDS-SPSP and the NNIPPS criteria of PSP in a largeretrospective series of pathologically confirmed PSPcases and appropriate disease controls, which had beenobserved by neurologists in tertiary neurologicalcenters.

Patients and Methods

Protocol Approvals and Patient Consents

Approval for this work was obtained from the localinstitutional review boards and the ethics committee of thePhilipps University (Marburg, Germany). Before death, alldonors gave written informed consent for the use of theirbrain tissue and medical records for research purposes.

Identification of Cases and Controls

We interrogated the patient registries of four Euro-pean brain banks with expertise in neurodegenerativedisorders to identify patients with the pathological di-agnosis of PSP, brainstem predominant Lewy body

(LB) disease with clinical history of Parkinsons’s die-ase (PD), MSA with predominant parkinsonism(MSA-P), and corticobasal degeneration (CBD). Thesebrain banks were the following: (1) Center for Neuro-pathology and Prion Research, Ludwig-Maximilians-University (Munich, Germany); MRC London Neuro-degenerative Diseases Brain Bank, King’s College(London, UK); Netherlands Brain Bank, Amsterdam incollaboration with the Department of Neurology,Erasmus Medical Center (Rotterdam, The Nether-lands); and Neurological Tissue Bank of the Biobanc-Hospital Clinic/Institut d’Investigacio Biomedica Au-gust Pi i Sunyer (Barcelona, Spain) in collaborationwith the Neurology Department of the HospitalClinic.

Cases that have entered these brain banks originatedfrom multiple tertiary neurological hospitals. There-fore, they are not filtered by individual academic cen-ters with a specialized clinical focus. Diagnosis wasbased on currently accepted criteria for the pathologi-cal diagnosis of PSP,2,6 PD,7 MSA,8 and CBD.9 Caseswere randomly selected for our study from the inven-tories of the participating brain banks. Cases withinsufficient clinical data were subsequently excluded.

Data Collection

Retrospective chart review of PSP cases and controlswas performed to extract clinical information gatheredbetween 1973 and 2008. We carefully analyzedpatient files to identify standardized information aboutdemography and the following clinical features: pos-tural instability; falls; supranuclear vertical gaze palsy;abnormal saccades or pursuits; nonspecific visual

symptoms; bradykinesia; axial rigidity; limb rigidity;tremor; dysarthria; dysphagia; cognitive dysfunction;

frontal lobe dysfunction; frontal physical signs; per-sonality change; social dysfunction; executive dysfunc-

tion; frontal behavior; axial dystonia; limb dystonia;

apraxia of limb(s); asymmetric onset; persistent asym-

metry; alien limb phenomenon; cortical sensory loss;

myoclonus; progressive nonfluent aphasia; apraxia of

speech; hallucinations unrelated to dopaminergic ther-

apy; dysautonomia; cortical dementia of Alzheimer’s

type; and response to levodopa. Absence or presence

of the above-listed symptoms, as well as year of onset,

was recorded. Features that could not be abstracted

from the files were recorded as “not available” and

excluded from analysis. Based upon these clinical

data, the fulfilment of the NINDS-SPSP and NNIPPS

criteria was retrospectively verified for each of the first

8 years after disease onset and for the final ante mor-

tem record. The NINDS-SPSP criteria for possible and

probable PSP as well as a combination of both,

referred to as NINDS-SPSP “total,” were separately

P S P D I A G N O S T I C C R I T E R I A

Movement Disorders, Vol. 28, No. 4, 2013 505

analyzed. NINDS-SPSP total includes all patients that

either fulfill the NINDS-SPSP criteria for possible or

probable PSP.

Statistical Analysis

Data are presented as the mean 6 standard error ofthe mean (SEM). A P value <0.05 was assumed to bestatistically significant. Normal parametric values were

compared by analysis of variance, followed by thepost-hoc least significant difference test.

Results

We were able to collect detailed demographic and clini-caldataof98pathologicallyconfirmedPSPpatientsand46

TABLE 1. The National Institute of Neurological Disorders and Stroke and Society for Progressive Supranuclear Palsy(NINDS-SPSP) and the Neuroprotection and Natural History in Parkinson Plus Syndromes (NNIPPS) diagnostic criteria

Inclusion Criteria Exclusion Criteria

NINDS-SPSP All of the following: Any of the following:� Age at disease onset �40 years; � Recent history of encephalitis� Gradually progressive disorder � Alien limb syndrome

Probable A. Vertical supranuclear palsy � Cortical sensory deficitsAnd � Focal frontal or temporoparietal atrophyB. Postural instability with falls within 1 year of disease onset � Hallucinations or delusions unrelated to

dopaminergic therapy� Cortical dementia of Alzheimer’s type

Possible A. Vertical supranuclear palsy � Prominent, early cerebellar symptomsOr � Early unexplained dysautonomiaB. 1. Slowing of vertical saccades � Severe, asymmetric parkinsonian signs

� Neuroradiologic evidence of relevant structural abnormality� Whipple’s diesease

And2. Postural instability with falls within 1 year of disease onset

NNIPPS All of the following: Any of the following:� Supranuclear ophthalmoplegia � Idiopathic Parkinson’s disease� Postural instability or falls � Evidence of any other neurological disease(within 3 years from disease onset) that could explain signs� Akinetic-rigid syndrome � History of repeated strokes with stepwise� Disease duration (12 months to 8 years) progression of parkinsonian features� Age at disease onset �30 years � History of major stroke

� Any history of severe or repeated head injury� History of encephalitis� History of neuroleptic use for a prolonged period oftime or within the past 6 months� Street-drug–related parkinsonism� Significant other neurological disease on CT scan/MRI� Oculogyric crises� Signs of corticobasal degeneration� Signs of Lewy body disease� Cerebellar ataxia� Symptomatic autonomic dysfunction� Tremor at rest

TABLE 2. Demographic data of the postmortem verified PSP cases and controls

Cases Controls

Demographic PSP All PD MSA-P CBD

N 98 46 18 13 15Males/females 55:43 19:27 5:4 2:11 7:8Age of onset, years[mean 6 SEM (range)]

64.4 6 1.2 (42–91) 61.5 6 1.3 (45–75) 62,6 6 2.4 (45–72) 57.6 6 2.1 (47–75) 64.5 6 1.7 (54–74)

Age of death, years[mean 6 SEM (range)]

72.2 6 0.9 (51–93) 70.1 6 1.2 (56–84) 74.7 6 1.9 (61–84) 64.9 6 1.8 (56–77) 69.3 6 1.5 (62–78)

Disease duration, years[mean 6 SEM (range)]

7.6 6 0.4 (0.5–19) 9.6 6 0.8 (3–26) 13 6 1.5 (4–26) 8.2 6 0.6 (3–11) 6.2 6 0.6 (3–10)

R E S P O N D E K E T A L .

506 Movement Disorders, Vol. 28, No. 4, 2013

disease controls (PD, MSA-P, and CBD). Their demo-graphicdataaregiveninTable1.

Clinical diagnosis of PSP, as made by the treating physi-cians, was given correctly in only 19% at the first clinicalvisit and in 71% throughout the course of disease in the

postmortem validated PSP cases. Twenty-four percent ofthe controls were falsely diagnosed during their lifetime asPSP(PD,22.2%;MSA-P,15.4%;CBD,36.4%).

Sensitivity, specificity, PPV, and negative predictivevalue (NPV) of the NINDS-SPSP and NNIPPS criteria

FIG. 1. Accuracy of the NNIPPS criteria and the possible, probable, and total (i.e., either possible or probable) NINDS-SPSP criteria as function ofdisease duration, regarding sensitivity (A), specificity against all controls (B), specificity against PD (C), specificity against MSA-P (D), specificityagainst CBD (E), PPV (F), and NPV (G).

P S P D I A G N O S T I C C R I T E R I A

Movement Disorders, Vol. 28, No. 4, 2013 507

after retrospective application to our series of post-mortem validated PSP, as functions of disease dura-tion, are shown in Figure 1. Note that the decrease ofsensitivity of the NNIPPS criteria between 8 years ofdisease duration and the final record was caused bythe exclusion of patients with a survival time greaterthan 8 years from PSP diagnosis, according to theNNIPPS criteria. Overall, 35.4% of PSP cases in ourseries had a survival time greater than 8 years.

The lowest mean time from disease onset to the firstclinical diagnosis of PSP was achieved by the NINDS-SPSP criteria for probable PSP (2.8 6 0.2 years), fol-lowed by the NNIPPS criteria (3.1 6 0.3 years; notsignificantly different), whereas the time to diagnosisby the NINDS-SPSP criteria for possible PSP was sig-nificantly longer (4.4 6 0.4 years; P < 0.01 versusNINDS-SPSP probable and P < 0.05 versus NNIPPS).

Discussion

We have evaluated the diagnostic accuracy of theNINDS-SPSP and the NNIPPS criteria of PSP in a largeseries of pathologically confirmed PSP cases and diseasecontrols with other neurodegenerative parkinsinsonism.

Patients originated from multiple academic hospitalsand were not filtered by specialized clinics with a par-ticular scientific or clinical focus. The qualification ofthe examining physicians ranged from general neurol-ogists to consultants with particular expertise in move-ment disorders. Thus, the quality of clinical examina-tion in our cohort may be considered as representativefor the performance of the NINDS-SPSP and NNIPPScriteria in a tertiary medical care setting. Because ofthe retrospective nature of the study, the examiningphysicians did not receive any specific instructions inwhich features to assess and document. Consequently,the clinical data documented for each patient and eachyear varied with regard to its detailedness and focus.Likewise, the data obtained from clinical charts mighthave been incomplete. As in any pathological series,we cannot exclude a selection bias overrepresentingdiagnostically more challenging cases with unusualclinical course. Such a bias toward autopsy in clini-cally equivocal cases might explain the surprisinglyhigh rate of clinical diagnosis of PSP in neuropatho-logical diagnosed PD in this series (22.2%) and mighthave resulted in a lower specificity of the NINDS andNNIPPS criteria. However, the observed rate of non-concordant clinical diagnoses of PSP in neuropatho-logically proven MSA-P and CBD patients, as well asthe rate of incorrect divergent clinical diagnoses inpathologically proven PSP patients compares well withpreviously published clinicopathological series.3,10,11

To avoid these limitations, a long-term, prospectivestudy would be required, allowing standardized collec-

tion of the clinical information and the pathologicaldiagnosis of PSP patients and controls.

For the first time, we have verified the accuracy of theNNIPPS criteria5 in an independent cohort, providingannual evaluations of sensitivity, specificity, and PPVthroughout the course of disease. In our series ofpatients and controls, the sensitivity of the NNIPPS cri-teria did not significantly differ during the first 3 yearsof disease duration from NINDS-SPSP probable criteriaand yielded slightly lower specificity and PPV. Themodifications of the NNIPPS criteria that intended toincrease sensitivity were an extension of the limits forage of onset from �40 to �30 years and for time to pos-tural instability with falls from 1 to 3 years after diseaseonset. However, the youngest PSP case in our cohortshowed first clinical signs at age 42. Also, 61% of thePSP patients with a record of postural instability andfalls developed these in the first year after clinical dis-ease onset; only 18% developed them in the second orthird year; but in these, other features, especially ab-sence of supranuclear gaze palsy and presence of tremorat rest, prevented the diagnosis of PSP according to theNNIPPS criteria. Exclusion of patients with a diseasehistory longer than 8 years on the basis of the NNIPPScriteria also led to an exclusion of many PSP patientsfrom a positive diagnosis at the final record. In sum-mary, the NNIPPS criteria did not exceed the NINDS-SPSP criteria in our series of patients with regard to sen-sitivity, specificity, and PPV.

The NINDS-SPSP criteria for possible PSP were con-sidered to increase sensitivity at the expense of speci-ficity.1,3 In our series, these criteria yielded the lowestspecificity, sensitivity, and PPV. The unexpectedly lowsensitivity of the NINDS criteria for possible PSPresulted from an infrequent recording of abnormalsaccades: The presence or absence of saccadic abnor-malities and the time of onset was reported in only36% of PSP patients in this study. Because of the ret-rospective nature of our study, we cannot determinewhether this resulted from infrequent presence of thesymptom or, rather, a poor recognition or a lack ofdocumentation by the general neurologists withoutspecial training to depict such deficits. Still, we wouldasume that a careful examination of eye movementsand recording of vertical saccades would haveimproved the diagnostic accuracy of the clinical crite-ria for the diagnosis of PSP in our series.

Conclusion

Based on these data, we suggest that the totalNINDS-SPSP criteria, accepting both possible andprobable for diagnosis, yielded the highest sensitivityin this tertiary medical care setting and might thus bemost useful for routine clinical care. For scientific clin-ical trials, the NINDS-SPSP probable criteria might be

R E S P O N D E K E T A L .

508 Movement Disorders, Vol. 28, No. 4, 2013

preferred, posessing the highest PPV and allowing, atpresent, the earliest, most specific diagnosis, thusallowing us to include correct positives early on andto reliably exclude false positives.

Acknowledgments: The brain banks thank all brain donors andrelatives for their invaluable support of research.

References1. Litvan I, Agid Y, Jankovic J, et al. Accuracy of clinical criteria for

the diagnosis of progressive supranuclear palsy (Steele-Richardson-Olszewski syndrome). Neurology 1996;46:922–930.

2. Hauw JJ, Daniel SE, Dickson D, et al. Preliminary NINDS neuro-pathologic criteria for Steele-Richardson-Olszewski syndrome (pro-gressive supranuclear palsy). Neurology 1994;44:2015–2019.

3. Osaki Y, Ben-Shlomo Y, Lees AJ, et al. Accuracy of clinical diagnosisof progressive supranuclear palsy. Mov Disord 2004;19:181–189.

4. Collins SJ, Ahlskog JE, Parisi JE, Maraganore DM. Progressivesupranuclear palsy: neuropathologically based diagnostic clinicalcriteria. J Neurol Neurosurg Psychiatry 1995;58:167–173.

5. Bensimon G, Ludolph A, Agid Y, Vidailhet M, Payan C, LeighPN; NNIPPS Study Group. Riluzole treatment, survival, and diag-nostic criteria in Parkinson plus disorders: the NNIPPS study.Brain 2008;132:156–171.

6. Dickson DW, Ahmed Z, Algom AA, Tsuboi Y, Josephs KA. Neu-ropathology of variants of progressive supranuclear palsy. CurrOpin Neurol 2010;23:394–400.

7. Braak H, DelTredici K, Rub U, deVos RA, Jansen Steur EN, BraakE. Staging of brain pathology related to sporadic Parkinson’s dis-ease. Neurobiol Aging 2003;24:197–211.

8. Trojanowski JQ, Revesz T; Neuropathology Working Group onMSA. Proposed neuropathological criteria for the post mortem di-agnosis of multiple system atrophy. Neuropathol Appl Neurobiol2007;33:615–620.

9. Dickson DW, Bergeron C, Chin SS, et al. Office of Rare Diseasesneuropathologic criteria for corticobasal degeneration. J Neuropa-thol Exp Neurol 2002;61:935–946.

10. Litvan I, Goetz CG, Jankovic J, et al. What is the accuracy of theclinical diagnosis of multiple system atrophy? A clinicopathologicstudy. Arch Neurol 1997;54:937–944.

11. Boeve BF, Maraganore DM, Parisi JE. Pathologic heterogeneity inclinically diagnosed corticobasal degeneration. Neurology 1999;53:795–800.

P S P D I A G N O S T I C C R I T E R I A

Movement Disorders, Vol. 28, No. 4, 2013 509