The prevalence, correlation, and co-occurrence of neuropathology in old age: harmonisation of 12 measures across six community-based autopsy studies of dementia

Summary Background Population-based autopsy studies provide valuable insights into the causes of dementia but are limited by sample size and restriction to specific populations. Harmonisation across studies increases statistical power and allows meaningful comparisons between studies. We aimed to harmonise neuropathology measures across studies and assess the prevalence, correlation, and co-occurrence of neuropathologies in the ageing population. Methods We combined data from six community-based autopsy cohorts in the US and the UK in a coordinated cross-sectional analysis. Among all decedents aged 80 years or older, we assessed 12 neuropathologies known to be associated with dementia: arteriolosclerosis, atherosclerosis, macroinfarcts, microinfarcts, lacunes, cerebral amyloid angiopathy, Braak neurofibrillary tangle stage, Consortium to Establish a Registry for Alzheimer's disease (CERAD) diffuse plaque score, CERAD neuritic plaque score, hippocampal sclerosis, limbic-predominant age-related TDP-43 encephalopathy neuropathologic change (LATE-NC), and Lewy body pathology. We divided measures into three groups describing level of confidence (low, moderate, and high) in harmonisation. We described the prevalence, correlations, and co-occurrence of neuropathologies. Findings The cohorts included 4354 decedents aged 80 years or older with autopsy data. All cohorts included more women than men, with the exception of one study that only included men, and all cohorts included decedents at older ages (range of mean age at death across cohorts 88·0–91·6 years). Measures of Alzheimer's disease neuropathological change, Braak stage and CERAD scores, were in the high confidence category, whereas measures of vascular neuropathologies were in the low (arterioloscerosis, atherosclerosis, cerebral amyloid angiopathy, and lacunes) or moderate (macroinfarcts and microinfarcts) categories. Neuropathology prevalence and co-occurrence was high (2443 [91%] of 2695 participants had more than one of six key neuropathologies and 1106 [41%] of 2695 had three or more). Co-occurrence was strongly but not deterministically associated with dementia status. Vascular and Alzheimer's disease features clustered separately in correlation analyses, and LATE-NC had moderate associations with Alzheimer's disease measures (eg, Braak stage ρ=0·31 [95% CI 0·20–0·42]). Interpretation Higher variability and more inconsistency in the measurement of vascular neuropathologies compared with the measurement of Alzheimer's disease neuropathological change suggests the development of new frameworks for the measurement of vascular neuropathologies might be helpful. Results highlight the complexity and multi-morbidity of the brain pathologies that underlie dementia in older adults and suggest that prevention efforts and treatments should be multifaceted. Funding Gates Ventures.


Appendix 1: Cohort descriptions
Adult Changes in Thought (ACT) Study: ACT was started in 1994, and recruits participants who are members of the Kaiser Permanente Washington health plan (previously Group Health), who are over 65, do not have dementia, do not reside in a nursing home, and who have been enrolled in the health plan for at least two years. Participants are roughly reflective of the population of King County, Washington. In addition to the initial cohort, an expansion cohort, and an ongoing replacement cohort were added. Ongoing recruitment targets a stable cohort of ~2,000 participants. About a third of participants who have died have consented to brain autopsy.

Framingham Heart Study (FHS):
Participants of the Framingham Heart Study (FHS) were originally recruited in 1948 based via mailed invitation letters based on random sampling of 2/3 of families with members aged 30-59 living in Framingham MA, USA. Additional cohorts, including the Offspring and Third Generation cohorts have created and added to the intergenerational nature of the study. The autopsy program began in 1997 and is voluntary. Autopsy data are available for selected participants in the Original and Offspring cohorts.

Cambridge City over-75s Cohort Study (CC75C):
The Cambridge City over-75s Cohort Study targeted all individuals 75 years and old who were registered at selected general medical practices in Cambridge, England, UK. General practices were selected to be geographically and socially representative of the city. The original survey in 1985 had a response rate of 95%, and individuals have been followed for 28 years. Although recruitment for the brain autopsy program initially focused on those with lower cognition, subsequent efforts were made to add individuals with normal cognitive functioning and the distribution of Mini-Mental State Examination (MMSE) scores in the autopsy sample closely matches the distribution in the full cohort at baseline.

Cognitive Functioning and Ageing Studies (CFAS):
The CFAS study had five centers in England and Wales (Cambridgeshire, Gwynedd, Newcastle upon Tyne, Nottingham, and Oxford). Random samples of 2500 participants aged 64 years and older were sampled. An additional 5200 individuals were included from a sixth center based in Liverpool. The study included two waves (four in Liverpool), and participants were invited to participant in a brain donation program.

Honolulu-Asia Aging Study (HAAS):
The Honolulu-Asia Aging Study (HAAS; baseline 1990) includes 80% surviving members of the existing Honolulu Heart Program (HHP), which started in 1965 as a community-based cohort study of cardiovascular disease. The original cohort included all Japanese-American men born between 1900 and 1919 and listed on the Honolulu Selective Service rolls for World War II and still living on Oahu, Hawaii at baseline. The autopsy study was started in 1992, when an invitation to participate was offered to all men regardless of dementia diagnosis. Analysis of brain pathology followed a standardized protocol.

The Religious Orders Study and Memory and Aging Project (ROSMAP):
The Religious Orders Study (ROS) began in 1994 and enrolls nuns, priests, and brothers from across the US. The Memory and Aging Project started in 1997 and includes lay persons from across northeastern Illinois recruited from retirement communities. Both studies are ongoing with continuing recruitment and require consent to autopsy at study entry. Both studies also have biannual data collection and protocols and data collection Appendix 2: Dementia Ascertainment by Cohort Adult Changes in Thought (ACT) Study: All participants were assessed with the Cognitive Abilities Screening Instrument (CASI) at biennial study visits. Individuals who scored lower than 86 on the CASI or those who were referred due to staff or family concerns underwent comprehensive neuropsychological testing and detailed physical and neurological examinations. Dementia diagnoses were made after thorough medical records review and discussion of examinations via consensus conference. Diagnoses were based on the Diagnostic and Statistical Manual of Mental Disorders (DSM)-IV criteria.
Framingham Heart Study (FHS): At every biannual study visit, participants are administered the Mini-Mental State Examination (MMSE). Participants are flagged for further testing in a number of different ways: a score < 23 for anyone, <24 for persons with a high school education, < 26 for college-educated participants, decline of 3 points from a prior exam, decline of 5 points from person best score. Participants may also be flagged by themselves, their family or their physician. If flagged, participants receive a full neuropsychological battery and a neuropsychologist assigns a severity score based on this testing. Participants flagged in this stage then receive a full neurologic exam, and those who are flagged in this stage are considered by a dementia review panel. The review panel (consisting of a neurologist and neuropsychologist) makes a definition of dementia based on DSM-IV criteria.
Cambridge City over-75s Cohort Study (CC75C): All study participants initially completed the MMSE. Those with low to moderate scores (0-23) and ¼ of participants with mild or minimal cognitive impairment (24-25) received more in-depth interviews via the Cambridge Diagnostic Examination for the Elderly (CAMDEX), which was administered by a psychiatrist and includes a proxy informant interview. At death, two clinicians agreed on a consensus diagnosis consistent with DSM-IV criteria following review of all interviews, proxy informant reports, death certificates and retrospective interviews with relatives following the participant's death.

Cognitive Functioning and Ageing Studies (CFAS):
The screening interview included basic questions on demographics, health, functional impairment, and cognition (the MMSE). All individuals who were either suspected of having dementia or who had an MMSE less than or equal to 21, ½ to 1/10 of individuals who were not suspected of dementia and had MMSE scores above 21 were invited for a more comprehensive assessment using the Geriatric Mental State (GMS) examination. The information in the GMS examination was used in the Automated Geriatric Examination for Computer Assisted Taxonomy (AGECAT) algorithm. The GMS-AGECAT algorithm has been validated internationally against DSM-III-R criteria. When missing data precluded the use of the GMS-AGECAT algorithm, diagnosis was given by a clinician after review of all available information.

Honolulu-Asia Aging Study (HAAS):
At every examination, the 100-point Cognitive Abilities Screening Instrument (CASI) was used to assess cognitive functioning. At baseline, individuals with poor CASI score (<74) as well as stratified samples of participants with higher scores received a full dementia evaluation, including proxy interview. At later visits, only those with low CASI scores received a full dementia evaluation. When it was not possible to complete a full evaluation, CASI score of <65 was used to signal definite cognitive impairment, 65-73 marginal cognitive impairment, 74-82 low-normal and greater than 82 normal. The more comprehensive dementia evaluations included a neurological examination, -neuropsychological testing, and an informant interview. A consensus diagnosis of dementia was based on DSM III-R criteria.

The Religious Orders Study and Memory and Aging Project (ROSMAP):
At each annual examination, all participants underwent a comprehensive clinical assessment and neuropsychological battery. Educationstratified cutoffs for 11 cognitive tests covering 5 cognitive domains were developed and used to assess cognitive impairment. These impairment ratings along with other cognitive tests scores, basic demographics, and information on sensory and motor deficits or difficulties during cognitive testing were reviewed by neuropsychologists to come to a decision regarding the presence of clinical impairment in each of the 5 cognitive domains. A clinician then reviewed decisions on impairment status along with all other relevant clinical information to come to a clinical judgement on the presence of dementia. At the time of death, all relevant clinical information over time (but no information based on autopsy assessments) was re-reviewed to come to a final decision on clinical status at death.

Figure S1. Missing data on neuropathologies for each cohort
Missing data is relatively low for most measures across the majority of cohorts, with some exceptions. Microinfarcts, macroinfarcts, and cerebral amyloid angiopathy were available in a subset of individuals in HAAS. LATE-NC stage was only assessed in a subset of individuals in FHS.

Appendix 3: Dichotomization of six key neuropathologies included in UpSet plots
Braak Stage: Present corresponds to Braak stage of V or VI CERAD neuritic plaque score: Present corresponds to CERAD score of moderate or severe Macroinfarcts: Present corresponds to any macroinfarcts present Microinfarcts: Present corresponds to any microinfarcts present LATE-NC stage: Present corresponds to LATE-NC stage of 2 or greater Lewy bodies: Present corresponds to any Lewy body pathology present

Appendix 4: Methodological details on analyses to account for selection bias in the ACT cohort
Those who choose to consent to autopsy may be different from those who do not consent. While we had access to data on those who consented, the inferences and conclusions we want to make are about the entire cohort. In all cohorts except ROSMAP (which required consent to autopsy for study entry), only a subset of participants agreed to autopsy. To test the potential effects of this bias, we used inverse probability weighting to account for this potential source of bias in the ACT cohort, which contained the information necessary to construct inverse probability weights.
In ACT, we first identified variables that could plausibly be associated with consent to autopsy, and which could influence our findings and conclusions. The variables we considered were: -Age (& Age 2 to capture nonlinear associations) -Gender -Education (yrs) -Dementia -Self-reported Hispanic ethnicity -Nursing home residence (yes/no) -Marital status (never married/married/separated or divorced/other) -Smoking status (never smoker/past smoker/current smoker) -Stroke (self-reported) -Heart disease (myocardial infarction, angina, CABG, angioplasty) (self-reported) -Diabetes (self-reported) -Center for epidemiologic studies -depression scale (CESD) (per point) We estimated a logistic regression to predict consent to autopsy among individuals who had died (and therefore were eligible to be able to contribute autopsy data) using the above variables. We used the missing indicator method to account for missing data in the predictors of consent to autopsy. 1 Although this method can lead to bias when trying to estimate the effect of an exposure on an outcome, because our primary goal is prediction, we expect meaningful error to be minimal. We estimated weights in those who consented to autopsy as: The mean of the weights was 1.00, with a range from 0.36-8.23. We evaluated the performance of the weights by contrast the balance between the full sample and the autopsy sample using standardized mean differences for continuous variables and differences in proportions for binary variables. We assessed differences both before and after weighting. Weighting achieved balance on all of the predictors considered.

Figure S2. Assessment of balance on key predictors before and after inverse probability weighting to account for selection into the autopsy cohort in the ACT sample
Appendix Large vessel cerebral atherosclerosis rating by visual inspection at the Circle of Willis at the base of the brain. Included evaluation of the vertebral, basilar, posterior cerebral, middle cerebral, and anterior cerebral arteries and their proximal branches. ACT Atherosclerosis was identified grossly by neuropathologists. This was defined as "mild" when restricted to branch points in the circle of Willis, "moderate" when also in other regions at the base of the brain, and "severe" when present on the convexity of the cerebrum) CFAS Gross appearance and degree of atherosclerosis of large vessels. Graded as none = 0, mild = 1, moderate = 2, severe = 3.

CC75C
Gross appearance and degree of atherosclerosis of large vessels. Graded as none = 0, mild = 1, moderate = 2, severe = 3. Semi-quantitative scores (0-5 range) were generated in the frontal cortex, superior temporal cortex, entorhinal cortex, hippocampus CA1, and inferior parietal cortex. Scores of 0 were labeled none, a score of 1-3 was labeled mild, a score of 4 was labeled moderate, and a score of 5 was labeled severe.  Severity of vascular amyloid deposits in the brain parenchyma/meninges of frontal, temporal, parietal, and occipital cortices respectively. The pathology is graded as none = 0, sparse (one or two affected vessels per section) =1, moderate (several vessels per section) = 3 and severe (many affected vessels per section) = 5.

CC75C
Overall average cortical score for CAA (based on maximum scores for either parenchymal or meningeal CAA for each of the four cortical areas) converted back to categorical measure (similar to Love et al. 2014). Severity of vascular amyloid deposits in the brain parenchyma/meninges of frontal, temporal, parietal, and occipital cortices respectively. The pathology is graded as none = 0, sparse (one or two affected vessels per section) =1, moderate (several vessels per section) = 3 and severe (many affected vessels per section) = 5. Framingham Assessment addressed the question: Is (Cerebral) Amyloid angiopathy present? Scored none, mild, moderate, and severe HAAS Neocortical CAA rating: CAA grades were assigned using a system based on the number of CAA-positive parenchymal vessels per area of neocortex. Diffuse Plaques RUSH Semi-quantitative scores (0-5 range) were generated in the frontal cortex, superior temporal cortex, entorhinal cortex, hippocampus CA1, and inferior parietal cortex. Scores of 0 were labeled none, a score of 1-3 was labeled mild, a score of 4 was labeled moderate, and a score of 5 was labeled severe. ACT Variable not available. CFAS By standardized methods (Mirra et al., 1991). Maximum cortical diffuse plaque score. Based on severity of amyloid beta protein-reactive plaque deposits in the frontal, temporal, parietal and occipital cortices respectively. The pathology is graded as none = 0, sparse (one or deposits per section) = 1, moderate (several deposits per section) = 2 and severe (many deposits per section) = 3. Plaque density is referenced to images in CERAD Handbook. CC75C By standardized methods (Mirra et al., 1991). Maximum cortical diffuse plaque score. Based on severity of amyloid beta protein-reactive plaque deposits in the frontal, temporal, parietal and occipital cortices respectively. The pathology is graded as none = 0, sparse (one or deposits per section) = 1, moderate (several deposits per section) = 2 and severe (many deposits per section) = 3. Plaque density is referenced to images in CERAD Handbook. Framingham CERAD Diffuse plaques (plaques with non-compact amyloid and no apparent dystrophic neurites) (through 2014 Presence or absence of microinfarcts. Framingham Assessment addressed the question: Are one or more cortical, microinfarcts (including "granular atrophy") present?

HAAS
Microinfarcts were defined as temporally remote (judged older than 1 month) microscopic (not seen at gross examination) foci of neuronal loss and gliosis, or of focal leukoencephalopathy when the lesion occurred in white matter. Total microinfarcts were typically counted for several brain regions on standard H&E section representing 0.5-1.5 square cm of tissue. The microinfarct counts used for these analyses were the total number identified on sections from the isocortex (8 sections), caudate (n = 2), putamen (n = 2), globus pallidus (n = 2), thalamus (n = 2), hippocampus (n = 2), nucleus basalis (n = 2), amygdala (n = 2), brainstem (n = 1), pons (n = 1), and cerebellum (n = 2 *Give information separately for exposed and unexposed groups. Note: An Explanation and Elaboration article discusses each checklist item and gives methodological background and published examples of transparent reporting. The STROBE checklist is best used in conjunction with this article (freely available on the Web sites of PLoS Medicine at http://www.plosmedicine.org/, Annals of Internal Medicine at http://www.annals.org/, and Epidemiology at http://www.epidem.com/). Information on the STROBE Initiative is available at www.strobe-statement.org.