Staging of Sporadic Parkinson Disease-Related a-synuclein Pathology: Inter- and Intra-Rater Reliability

J Neuropathol Exp Neurol Copyright Ó 2005 by the American Association of Neuropathologists, Inc. Vol. 64, No. 7 July 2005 pp. 623 628 ORIGINAL ARTICLE Staging of Sporadic Parkinson Disease-Related a-synuclein Pathology: Inter- and Intra-Rater Reliability Christian M. Müller, PhD, Rob A. I. de Vos, MD, Claude-Alain Maurage, MD, Dietmar R. Thal, MD, Markus Tolnay, MD, and Heiko Braak, MD Abstract Sporadic Parkinson disease (spd) is characterized by a-synuclein (a-syn) inclusions. The distribution of such inclusions appears to relate to disease progression and severity. We propose and test a simple staging protocol based on the presence of a-syn immunoreactivity in 5 paraffin sections that, taken together, contain up to 8 vulnerable brain regions. Six stages of a-syn pathology reminiscent for spd are defined based on the presence or absence of inclusions in the assessed sections. Six observers from 5 different institutions rated 21 cases on the basis of written instructions only. The agreement of the raters was highly significant with a mean error below one stage. Both inter- and intra-rater reliability were also substantial to almost perfect as analyzed by paired comparison between all raters. We propose that the staging procedure for a-syn pathology is suitable for application in routine neuropathology and brain banking. Clearly defined stages of a- synpathology might aid the comparability between studies and also help to distinguish spd from other synucleinopathies. Key Words: Brain banking, Lewy bodies, Lewy neurites, Parkinson disease, Routine pathology, Staging method, Synucleinopathy. INTRODUCTION Sporadic Parkinson disease (spd), the second most common neurodegenerative disorder of the human nervous system, affects multiple brain regions. The clinical hallmarks of spd are olfactory dysfunction, autonomic and motor dysfunction, and cognitive decline. Neuronal degeneration is paralleled by the occurrence of almost dissoluble aggregates in neuronal somata (Lewy bodies [LBs]) and processes (Lewy neurites [LNs]) that contain a-synuclein (a-syn), ubiquitin, and phosphorylated neurofilaments (1). From the Institute for Clinical Neuroanatomy (CMM, HB), J. W. Goethe University Clinic, Frankfurt/M., Germany; Laboratorium Pathologie Oost Nederland (RAIvD), Enschede, The Netherlands; Hôpital Roger Salengro (C-AM), CHRU de Lille, Lille, France; Department of Neuropathology (DRT), University of Bonn, Bonn, Germany; and Institute of Pathology (MT), Department of Neuropathology, University Hospital, Basel, Switzerland. Send correspondence and reprint requests to: Dr. Christian M. Müller, J. W. Goethe University Clinic, Institute for Clinical Neuroanatomy, Theodor Stern Kai 7/Hs. 27A, D-60590 Frankfurt/M., Germany; E-mail: ch. mueller@med.uni-frankfurt.de Supported by the Deutsche Forschungsgemeinschaft (DFG, Br 317-17-2). Besides the presence of a-syn-positive LBs and LNs in the substantia nigra (SN) of clinically diagnosed spd cases, such pathology is also observed in extranigral structures. Patterns of a-syn aggregates not including the SN appear to reflect early stages of spd or might correspond to other related disorders with Lewy bodies (2). Known disorders also characterized by the occurrence of a-syn inclusions are dementia with Lewy bodies (DLB), pure autonomic failure (PAF), multiple system atrophy (MSA), and neurodegeneration with brain iron aggregations type I (NBIA-I, Hallervorden-Spatz) disease (3). Based on observations of postmortem tissue from diagnosed spd cases as well as from persons without clinical signs of spd, a successive propagation of a-syn-pathology from the medulla oblongata and olfactory bulb to telencephalic structures in end-stage spd has been proposed (4, 5). This classification helps to identify putatively preclinical stages of spd (6) and also aids the comparability of studies on postmortem tissue of spd-patients by defining the progression or severity of the disease pathology. A similar classification has proved to be useful over the last decades with respect to Alzheimer disease (7). The present study is aimed at introducing a simple staging procedure for a-syn pathology based on a minimum of tissue samples processed by routine neuropathological methods, that is, paraffin sections immunostained using an automated immunostainer and a commercially available a-synantibody. By comparing the ratings from 2 experienced observers and four observers without prior experience in the staging of a-syn pathology, we address the reliability of such a staging procedure for routine neuropathology and brain banking purposes. MATERIALS AND METHODS The study is based on tissue samples from brains of 21 subjects (9 females, 12 males, mean age 76.5 6 6.7 years) from the Department of Pathology in Enschede, The Netherlands. All cases were selected from a larger pool of pathological material based on the following criteria essential for the purpose of the study: i) availability of all necessary brain structures (see below); ii) the presence (spd cases) or absence (controls) of a-syn immunoreactivity (a-syn-ir) in different brain regions indicative of different stages of spd (4); and iii) absence of additional signs of neuropathology; for example, Alzheimer disease higher than neurofibrillary stage J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 623

Müller et al J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 III (8), brain hemorrhage or a known history of neurological disorders other than PD. The postmortem delay to autopsy ranged from 9 to 70 hours, with an average delay of 24 hours. As shown in Table 1, 10 of the patients (cases 11 13 and 15 21) had a documented diagnosis of Parkinson disease dating from 2 to 20 years prior to death, while the clinical records of the remaining cases did not mention such a diagnosis. TABLE 1. Summary of Available Data for the Cases Studied Case Sex Age NFT b-amyloid PD PD Duration 1 f 89 III 3 2 m 85 I 2 3 f 75 II 1 4 f 71 II 2 5 m 81 III 1 6 f 79 II 2 7 f 71 III 2 8 m 83 II 0 9 m 77 I 1 10 m 67 II 1 11 m 76 II 0 + 6 years 12 m 87 I 1 + 3 years 13 m 65 I 0 + nd 14 f 68 III 3 15 m 77 I nd + 12 years 16 m 77 II 2 + 2 years 17 f 78 III nd +. 4 years 18 m 78 II 1 + 10 years 19 m 82 II 3 +. 5 years 20 f 69 III nd + 20 years 21 f 71 II 2 + 18 years Cases are ordered by ascending modal of the stages assigned during the rating (see Table 3). NFT, neurofibrillary tangle stage; b-amyloid, b-amyloid stage; PD, clinical diagnosis for Parkinson disease; nd, not determined. Tissue Preparation Brains were fixed in 4% aqueous formaldehyde solution and subsequently processed as follows: Tissue samples were taken from i) the medulla oblongata at the level of the dorsal IX/X motor nucleus and adjoining intermediate reticular zone, ii) the coeruleus-subcoeruleus complex, iii) substantia nigra, iv) the uncal portion of the hippocampal formation, including the entorhinal cortex and anteromedial temporal mesocortex, and v) the neocortex covering the first temporal gyrus. A detailed and illustrated description of the brain regions utilized has been published in an earlier report (4). Tissue blocks were embedded in paraffin and cut at 10 mm. Tissue sections were immunostained using standard neuropathological routines and an automatic immunostainer (LV-1, Lab Vision Corp., Fremont, CA). Briefly, after deparaffination, sections were pretreated with formic acid (15 min), rinsed in distilled water, and autoclaved for 20 minutes in citrate buffer to enhance accessibility of the antibody (9). Sections then were incubated overnight at 37 C with a monoclonal antibody directed against a-syn (NCL-L-ASYN, Novocastra, Newcastle-upon-Tyne, UK; 1:60), and further processed for avidin-biotin-immunocytochemistry with diaminobenzidine as chromogen. Finally, the sections were counterstained with hematoxylin and coverslipped. Cases were selected from the available pool of samples to obtain a mean of 3 cases from each stage (stages 1 6) and controls, albeit not at identical frequency to reduce possible influences of rater expectancy during the staging procedure. The selected cases were anonymized using 3-digit random numbers. To further reduce a possible influence of expectation by raters with respect to the frequency of certain stages, cases were randomly separated into 2 groups of 10 and 11 cases, which were shipped to the raters with a time delay of at least 3 weeks. Based on the initial selection, the following frequencies of stages were present in the entire collection: controls (stage 0), n = 2; stage 1, n = 4; stage 2, n = 1; stage 3, n = 3; stage 4. n = 3; stage 5, n = 5; stage 6, n = 3. The 2 groups of tissue sections rotated among the 6 raters. Each rater had 7 days to perform the staging and to forward the samples to the next observer. At the end of the study, all raters had performed the staging on identical tissue sections twice. After completion of the first cycle, reanonymization was performed by assigning new random numbers for each case. Instructions to Raters All raters were supplied with written directives on the definition of each stage in the English language and a standardized form to protocol their observations. One of 4 possible rating options could be chosen for each analyzed brain structure: i) no a-syn immunoreactivity (a-syn-ir) present (2); ii) occasional a-syn-ir present (+); iii) moderate density of a-syn-ir present (++); and iv) extensive a-syn-ir present (+++). For the decision regarding a-syn-ir in the neocortex, raters were instructed to select minor pathology (+) when more than 4 LBs were confined to infragranular cortical layers, whereas moderate to extensive pathology (++, +++) was defined by the presence of LBs and LNs also in supragranular cortical layers. Raters were instructed to analyze the following brain structures: i) the dorsal IX/X motor nucleus; ii) intermediate reticular zone; iii) locus coeruleus; iv) caudal raphe nuclei/magnocellular portions of the reticular formation (if present); v) substantia nigra (pars compacta); vi) second sector of the Ammon s horn (CA2); vii) anteromedial temporal mesocortex covering the rhinal or collateral sulcus and adjoining neocortex; and viii) the neocortex of the first temporal gyrus. Finally, raters were requested to assign a stage based on the criteria summarized in Table 2. Statistical Analysis Consensus agreement of all raters was assessed using the statistics for nominal scale agreement introduced by Fleiss (10). To evaluate inter- and intra-rater reliability, we used the kappastatistics of Cohen (11, 12) with linear weighting. Weighted kappa (Equation 1) was calculated from the weighted observer agreement (Equation 2) and the weighted expected agreement (Equation 3). Linear weighting for observed frequencies f ij (Equation 4) was applied, with g being the number of stages (g = 6) and i and j the stages assigned by the first and second 624 q 2005 American Association of Neuropathologists, Inc.

J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 Reliability of Parkinson Disease Staging TABLE 2. Definition of Stages for a-synuclein Pathology According to a-syn Immunoreactivity (a-syn-ir) in Different Brain Regions Stage dmx or irz LC raphe/rf SN CA2 meso-cx neo-cx 0 1 $+ 2 $+ $+ 3 $+ $+ $+ $+ 4 $+ $+ $+ $+ $+ $+ 5 $+ $+ $+ $+ $+ $+ =+ 6 $+ $+ $+ $+ $+ $+ $++ Note that the severity of a-syn-ir is only essential for differentiating stages 5 and 6.dmX, dorsal IX/X motor nucleus; irz, intermediate reticular zone; LC, locus coeruleus; raphe/rf, caudal raphe nuclei/reticular formation; SN, substantia nigra; CA2, second sector of the Ammon s horn; cx, cortex. rater, respectively, in a given inter-rater-comparison. For intrarater comparison, i and j were the stages assigned in the first and second rating by a given rater. kw ¼ PoðwÞÿPeðwÞ 1 ÿ PeðwÞ PoðwÞ ¼ 1 n PeðwÞ ¼ 1 n2 i¼1 i¼1 j¼1 j¼1 w ij f ij w ij r i c j ji ÿ jj w ij ¼ 1 ÿ g ÿ 1 Weighted kappa values can vary from 0.00, indicating no better agreement than chance, to 1.00, indicating perfect agreement. The strength of the agreement was regarded as substantial for 0.61 to 0.80 and almost perfect for 0.81 to 1.00 (11). We also calculated the average error of each observer from the differences of each staging to the mean stage assigned by all raters. This value could vary between 0.00 when no deviation was observed to a theoretical maximum of 6.00 when, for example, a case generally rated as stage 6 was diagnosed as stage 0 by one rater. RESULTS Consensus Agreement of Different Raters Table 3 shows the frequency with which the 21 cases were rated by the 6 raters in 2 rating sessions (n = 12 ratings per case). Cases are ordered by ascending modal of rated stages for clarity. All cases with a clinical diagnosis of PD were rated at stage 4 or higher by all raters (compare Tables 1 and 3). In addition, one case for which no mention of PD was made in the clinical protocol (case 14) was rated at stage 5 by the majority of raters and at stage 4 by the remaining observers. As seen in Table 3, only 4 out of the total of 252 ratings differed by more TABLE 3. Frequency of Rated Stages for the 21 Cases Assessed By 6 Raters in 2 Independent Rating Sessions Rated Stages Cases 0 1 2 3 4 5 6 1 12 2 10 1 1 3 6 6 4 5 7 5 3 8 1 6 12 7 12 8 2 9 1 9 1 1 10 10 10 2 11 10 2 12 8 4 13 4 8 14 4 8 15 3 9 16 10 2 17 10 2 18 9 3 19 1 11 20 12 21 12 Cases are ordered by ascending modal of assigned stages. than one stage from the remaining ratings, and in all but one case (Table 3: case 3) a majority of raters decided for a common stage. In just 2 cases, less than two-thirds of the raters agreed on a given stage but rated at similar or identical frequency for the 2 adjacent stages 0 and 1 (cases 3 and 4 in Table 3). The high overall agreement of all raters is reflected in the statistical significance calculated from the logistic regression model proposed by Fleiss (10). The u-value of 46.067 corresponds to a statistical significance for inter-observer scale agreement of, 0.0001. Similar significances were reached upon testing the scale agreement for each of the 2 staging sessions separately (first rating: u = 23.456, p, 0.0001; second rating: u = 22.928, p, 0.0001). As will be discussed, the highest ambiguities were observed when raters had to distinguish between stages 0 and 1, and stages 4 and 5. Only in such cases was the modal stage selected by less than three-fourths of the raters. Compared to the stages originally assigned by one author during the selection of cases (see Materials and Methods), the modal stage assigned by all raters sometimes differed in these stages, whereas two cases with weak pathology (i.e. stage 1) were rated lower and 3 cases with advanced pathology (stages 4 and 5) were rated higher. Inter-Rater Reliability To address the issue of inter-rater agreement, we conducted pairwise tests of the ratings between all raters for both rating sessions (Tables 4, 5). There was substantial agreement (weighted kappa = 0.6 0.8) in all inter-rater comparisons q 2005 American Association of Neuropathologists, Inc. 625

Müller et al J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 TABLE 4. Pairwise Inter-Rater Reliability in the First Rating Session Raters DT MT CM RDV HB CMM 0.774 0.768 0.871 0.893 0.978 DT 0.761 0.819 0.838 0.797 MT 0.813 0.794 0.790 CM 0.896 0.894 RDV 0.916 The numerical values represent the weighted kappa for each combination of raters. Raters are designated by their initials. and even an almost perfect agreement (weighted kappa = 0.8 1.0) in the vast majority of pairwise comparisons (24/30). The inter-rater reliability test for the second staging session showed a slight improvement of the kappa values with respect to the first session (mean kappa-values: first session: 0.840, second session: 0.864). In addition, we determined the mean deviation of each examiner s rating with respect to the average stage assigned by all raters (Table 6). These rates of mistake varied between 0.21 and 0.40, being well below one stage class for all raters. Intra-Rater Reliability The two staging sessions yielded almost perfect agreement (weighted kappa = 0.8 to 1.0) in all raters (Table 7). Deviations between the assigned stage in the first and second staging exceeded one stage in only 4 out of 126 ratings. In an additional 21 ratings, a deviation of one stage was observed, while in the remaining 101 ratings identical stages were assigned to the rated cases. Severity of a-synuclein Pathology Whereas the staging protocol employed in the present study is based solely on the presence of a-syn-ir in selected brain regions, the raters also were asked to evaluate the density of a-syn pathology using a semiquantitative scale. An analysis of the relation of the severity of a-syn pathology and the stages assigned by raters to individual cases appears in Figure 1. To allow for a calculation of mean values, scales were converted to numerical values (2 = 0; + = 1; ++ = 2; +++ = 3). It is evident that the mean severity of a-syn pathology in brainstem nuclei is moderate at stages below stage 4 and increases to higher values at and beyond stage 4. Furthermore, lower densities of a-syn aggregates are observed within more central brain regions compared to those in brainstem nuclei and the substantia nigra up to stage 5. TABLE 5. Pairwise Inter-Rater Reliability in the Second Rating Session Raters DT MT CM RDV HB CMM 0.860 0.885 0.899 0.958 0.897 DT 0.829 0.800 0.857 0.800 MT 0.829 0.845 0.827 CM 0.897 0.878 RDV 0.895 TABLE 6. Average Error for Each Rater Relative to the Mean Staging By All Raters Raters CMM DT MT CM RDV HB Average error 0.24 0.37 0.40 0.29 0.21 0.24 DISCUSSION In the present study, we tested the reliability of criteria intended to characterize the distribution pattern of a-syn pathology related to spd. The definitions were based on an evaluation of stages in the progression of spd which has been deduced from thick (100 mm) sections stained for a-syn (4). Since the techniques for immunohistochemistry on thick sections are usually not available in routine neuropathology, we wanted to determine here whether the use of thin paraffin sections is suitable for distinguishing the stages of spd-related pathology. Although the group of raters included four observers without previous training in the applied staging protocol (i.e. those who rated exclusively on the basis of the supplied written instructions) inter-rater reliabilities revealed substantial (kappa values. 0.7) and, in the majority of cases, even almost perfect (kappa values. 0.8) agreement among the 6 raters. Similarly, intra-rater comparisons also revealed almost perfect agreement between 2 independent staging sessions. We therefore conclude that the proposed staging method is highly reliable even when inexperienced observers assess the rating in thin paraffin sections. An advantage of the proposed staging system is that it is based only on the presence rather than density of a-syn-ir within given brain structures. The only exception is the distinction between stages 5 and 6, where the extent of a-syn pathology has been included (Table 2). The overall lack of (semi-)quantitation allows objective assessment of stages even by inexperienced observes with adequate anatomical training. Addition of the primary sensory neocortex to the tissue samples would allow for an unequivocal definition for stage 6 (4); however, our study reveals an almost perfect inter-rater agreement for this distinction, thereby warranting adequate rating even without this additional sample. Thus, the proposed staging protocol is suitable for routine application in neuropathology, research, and brain banking. Even when considering deviations from a stereotypical, successive propagation of a-syn aggregation (13 15), the proposed rating scheme allows a defined case description. However, this study also shows that a direct correspondence between the severity of a-syn pathology and clinical symptoms typical of PD (bradykinesia, cogwheel rigidity, tremor, and postural instability) does not necessarily hold true in every instance. Within the group of cases rated here, for example, the clinical records TABLE 7. Intra-Rater Reliability for the 6 Raters Raters CMM DT MT CM RDV HB 0.960 0.944 0.874 0.838 0.958 0.922 Numerical values represent the weighted kappa upon comparison between the first and second rating session. 626 q 2005 American Association of Neuropathologists, Inc.

J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 Reliability of Parkinson Disease Staging FIGURE 1. Severity of a-synuclein immunoreactivity (a-syn-ir) in selected anatomical structures as determined from the semiquantitative assessment (2 = 0; + = 1; ++ = 2; +++ = 3) and related to the stages of a-syn-pathology. Note: the severity of a-syn lesions increases within each structure with increasing overall stage. dmx, dorsal IX/X motor nucleus; LC, locus coeruleus, raphe; SN, substantia nigra; CA2, second sector of the Ammon s horn; mx, mesocortex; tx, temporal cortex. of one case with severe a-syn pathology (case 14 in Table 1) failed to mention any characteristic PD symptoms. A similar dissociation between a-syn pathology and the clinical manifestation of PD has recently been described in a detailed study of 106 brains with a-syn-ir (13). To date, it is unclear whether such dissociations challenge the role of a-syn aggregates in the development of spd or whether such cases reflect putative preclinical stages in individuals with more subtle symptoms who only later develop spd. The latter assumption might be corroborated by longitudinal studies that show a higher incidence of PD in patients with presumed preclinical signs of PD, such as hyposmia (16, 17), REM-sleep disorder (18, 19), or autonomic dysfunctions (20); all of these disorders are believed to reflect disturbances in brain regions affected at stages 1 to 3 of a-syn pathology (4). An additional indication (though not a final proof) of a putative sequential involvement of brain regions with a-syn aggregates that begin in brainstem and anterior olfactory regions and proceed to more central structures (4) comes from the analysis of the severity of the a-syn pathology (Fig. 1): with increasing stages, the density of a-syn aggregates increases in brainstem structures and the lowest density of a-syn aggregates is consistently observed in structures differentiating a given stage from the lower one. Compared to the original description of the stages of spd-related a-syn pathology (4), the number of brain regions selected for the assessment of stages here was reduced to 5 tissue samples. For example, we excluded anterior olfactory structures, the basal forebrain nuclei, and the primary sensory cortex. With the exception of the latter, pathological a-synaggregation in all other regions has been shown to be paralleled by pathology in defined structures included here for each stage (4). Inasmuch as the staging system employed here took into account only the topographical distribution pattern of a-syn-ir observed in spd, an extension to additional brain regions has to be considered to also detect the possible presence of other synucleinopathies. For instance, a selective or predominant involvement of the amygdala has been described in some studies, which might be related to concomitant neurofibrillary tangle pathology (14, 15, 21, 22). Similarly, a-syn-ir has been shown to have a distinct distribution in multiple system atrophy (23, 24) and pure autonomic failure (25). The major ambiguity in the staging of our sample of cases occurred at the level of the distinction of controls from stage 1. The first signs of a-syn-aggregation are inclusions in neuronal processes, so-called Lewy neurites (5). These fine structures can easily be overlooked or mistaken for artefacts in thin paraffin sections. As such, in cases where detection of the earliest stages of spd-associated pathology is essential, one should consider using either thicker sections or at least multiple sections though the medulla oblongata at the dmxlevel. The second distinction, which appeared ambiguous in some cases, was the separation of stages 4/5 (Table 3; cases 12 14). Stage 5 is defined by the occasional presence of neocortical LBs (4). Again, the scarcity of this pathology within a large brain area increases the risk of missing the immunoreactive structures in paraffin sections. Decisions of individual raters deviated by more than one stage from the modal decision of the remaining raters in only 4 out of 252 ratings (Table 3). In 2 of these cases, the raters had commented their decision on the rating form, noting that they might have mistaken an artefact for immunostaining. One additional deviating rating also was attributable to mistaking artefacts for immunostaining, whereas the only significant underrating could only be explained by a failure to detect immunostaining in at least 2 brain regions. Unlike other staging procedures for synucleinopathies (26), the present approach does not rely on clinical data and, thus, can be applied to autopsy cases even where such data are unavailable. Since the proposed staging system also defines putatively preclinical PD-related pathology (5), it might complement approaches aimed at an early diagnosis of Parkinson disease that use imaging techniques or behavioral and sensory tests (27 31) by providing a postmortem evaluation that can be related to detailed clinical premortem examinations (32). Furthermore, the putative detection of presymptomatic cases of spd might aid the search for biomarkers paralleling or even preceding the evolution of a-syn aggregates visible in postmortem tissue. Besides helping to recognize different pathological states of the brain, the present staging system might also be of considerable value for defining controls (i.e. brains lacking a-syn-pathology) and for identifying deviations from the progression of the a-synucleinopathy seen in spd (13, 33). In summary, our study reveals that the evaluation of a-syn-pathology related to spd on the basis of a sequential involvement of defined brain regions can easily and reliably be reproduced by different raters and repeated observation by individual raters. Both of these features make the staging method suitable for brain banking and can considerably improve comparability of studies dealing with spd and related synucleinopathies. ACKNOWLEDGMENTS We are indebted to Mrs. A. te Vaanholt and B. Meseck- Selchow for expert preparation of sections and stainings, as q 2005 American Association of Neuropathologists, Inc. 627

Müller et al J Neuropathol Exp Neurol Volume 64, Number 7, July 2005 well as Dr. Kelly Del Tredici for critical discussions and reading the manuscript. REFERENCES 1. Dauer W, Przedborski S. Parkinson s disease: Mechanisms and models. Neuron 2003;39:889 90 2. Trojanowski JQ, Lee VMY. Parkinson s disease and related alphasynucleinopathies are brain amyloidoses. Ann NY Acad Sci 2003;991: 107 10 3. Galvin JE, Lee VMY, Trojanowski JQ. Synucleinopathies Clinical and pathological implications. Arch Neurol 2001;58:186 90 4. Braak H, Del Tredici K, Rüb U, et al. Staging of brain pathology related to sporadic Parkinson s disease. Neurobiol Aging 2003;24:197 211 5. Del Tredici K, Rüb U, de Vos RA, et al. Where does parkinson disease pathology begin in the brain? J Neuropathol Exp Neurol 2002;61:413 26 6. Mikolaenko I, Pletnikova O, Kawas CH, et al. Alpha-synuclein lesions in normal aging, Parkinson disease, and Alzheimer disease: Evidence from the Baltimore Longitudinal Study of Aging (BLSA). J Neuropathol Exp Neurol 2005;64:156 62 7. Hyman BT, Trojanowski JQ. Editorial on consensus recommendations for the postmortem diagnosis of Alzheimer disease from the National Institute on Aging and the Reagan Institute working group on diagnostic criteria for the neuropathological assessment of Alzheimer disease. J Neuropathol Exp Neurol 1997;56:1095 97 8. Braak H, Braak E. Neuropathological staging of Alzheimer-related changes. Acta Neuropathologica 1991;82:239 59 9. Hladik CL, White CL. Comparison of digestive enzyme and formic acid pretreatment for optimal immunohistochemical demonstration of alphasynuclein-immunoreactive cerebral cortical Lewy neurites. J Neuropathol Exp Neurol 2003;62:554 10. Fleiss JL. Measuring Nominal Scale Agreement Among Many Raters. Psych Bull 1971;76:378 82 11. Landis JR, Koch GG. Measurement of observer agreement for categorical data. Biometrics 1977;33:159 74 12. Landis JR, Koch GG. Application of hierarchical kappa-type statistics in assessment of majority agreement among multiple observers. Biometrics 1977;33:363 74 13. Parkkinen L, Kauppinen T, Pirttilä T, et al. Alpha-synuclein pathology does not predict extrapyramidal symptoms or dementia. Ann Neurol 2005;57: 82 91 14. Forman MS, Schmidt ML, Kasturi S, et al. Tau and alpha-synuclein pathology in amygdala of parkinsonism-dementia complex patients of Guam. Am J Pathol 2002;160:1725 31 15. Hamilton RL. Lewy bodies in Alzheimer s disease: A neuropathological review of 145 cases using alpha-synuclein immunohistochemistry. Brain Pathol 2000;10:378 84 16. Ponsen MM, Stoffers D, Booij J, et al. Idiopathic hyposmia as a preclinical sign of Parkinson s disease. Ann Neurol 2004;56:173 81 17. Berendse HW, Booij J, Francot CMJE, et al. Subclinical dopaminergic dysfunction in asymptomatic Parkinson s disease patients relatives with a decreased sense of smell. Ann Neurol 2001;50:34 41 18. Schenck CH, Bundlie SR, Mahowald MW. REM behavior disorder (RBD): Delayed emergence of parkinsonism and/or dementia in 65% of older men initially diagnosed with idiopathic RBD, and an analysis of the minimum & maximum tonic and/or phasic electromyographic abnormalities found during REM sleep. Sleep 2003;26:A316 19. Boeve B, Silber MH, Parisi JE, et al. Synucleinopathy pathology and REM sleep behavior disorder plus dementia or parkinsonism. Neurology 2003;61:40 45 20. Thaisetthawatkul P, Boeve BF, Benarroch EE, et al. Autonomic dysfunction in dementia with Lewy bodies. Neurology 2004;62:1804 9 21. Harding AJ, Stimson E, Henderson JM, et al. Clinical correlates of selective pathology in the amygdala of patients with Parkinson s disease. Brain 2002;125:2431 45 22. Lippa CF, Fujiwara H, Mann DMA, et al. Lewy bodies contain altered alpha-synuclein in brains of many familiar Alzheimer s disease patients with mutations in presenilin and amyloid precursor protein genes. Am J Pathol 1998;153:1365 70 23. Piao YS, Mori F, Hayashi S et al. alpha-synuclein pathology affecting Bergmann glia of the cerebellum in patients with alpha-synucleinopathies. Acta Neuropathologica 2003;105:403 9 24. Nishie M, Mori F, Fujiwara H, et al. Accumulation of phosphorylated alpha-synuclein in the brain and peripheral ganglia of patients with multiple system atrophy. Acta Neuropathologica 2004;107:292 98 25. Kaufmann H, Hague K, Perl D. Accumulation of alpha-synuclein in autonomic nerves in pure autonomic failure. Neurology 2001;56:980 81 26. Marui W, Iseki E, Nakai T, et al. Progression and staging of Lewy pathology in brains from patients with dementia with Lewy bodies. J Neurol Sci 2002;195:153 59 27. Yoshikawa K, Nakata Y, Yamada K, et al. Early pathological changes in the parkinsonian brain demonstrated by diffusion tensor MRI. J Neurol Neurosurg Psych 2004;75:481 84 28. Piccini P, Whone A. Functional brain imaging in the differential diagnosis of Parkinson s disease. Lancet Neurology 2004;3:284 90 29. Katzenschlager R, Lees AJ Olfaction and Parkinson s syndromes: Its role in differential diagnosis. Curr Op Neurol 2004;17:417 23 30. Przuntek H, Muller T, Riederer P. Diagnostic staging of Parkinson s disease: Conceptual aspects. J Neural Trans 2004;111:201 16 31. Lang AE, Obeso JA. Time to move beyond nigrostriatal dopamine deficiency in Parkinson s disease. Ann Neurol 2004;55:761 65 32. Braak H, Rüb U, Jansen Steur EN, et al. Cognitive status correlates with neuropathological stage in Parkinson disease. Neurology 2005;64: 1404 10 33. Parkkinen L, Soininen H, Alafuzoff I. Regional distribution of alphasynuclein pathology in unimpaired aging and Alzheimer disease. J Neuropathol Exp Neurol 2003;62:363 67 628 q 2005 American Association of Neuropathologists, Inc.