Plasma and cerebrospinal fluid ABeta42 for the differential diagnosis of Alzheimer's disease dementia in participants diagnosed with any dementia subtype in a specialist care setting

Michelle Kokkinou; Lucy C Beishon; Nadja Smailagic; Anna H Noel-Storr; Chris Hyde; Obioha Ukoumunne; Rosemary E Worrall; Anja Hayen; Meera Desai; Abhishekh Hulegar Ashok; Eleanor J Paul; Aikaterini Georgopoulou; Tiziana Casoli; Terry J Quinn; Craig W Ritchie

doi:10.1002/14651858.CD010945.pub2

. 2021 Feb 10;2021(2):CD010945. doi: 10.1002/14651858.CD010945.pub2

Plasma and cerebrospinal fluid ABeta42 for the differential diagnosis of Alzheimer's disease dementia in participants diagnosed with any dementia subtype in a specialist care setting

Michelle Kokkinou ¹, Lucy C Beishon ^2,^✉, Nadja Smailagic ³, Anna H Noel-Storr ⁴, Chris Hyde ⁵, Obioha Ukoumunne ⁶, Rosemary E Worrall ⁷, Anja Hayen ⁸, Meera Desai ⁹, Abhishekh Hulegar Ashok ^1,¹⁰, Eleanor J Paul ^1,¹⁵, Aikaterini Georgopoulou ¹¹, Tiziana Casoli ¹², Terry J Quinn ¹³, Craig W Ritchie ¹⁴

Editor: Cochrane Dementia and Cognitive Improvement Group

PMCID: PMC8078224 PMID: 33566374

Abstract

Background

Dementia is a syndrome that comprises many differing pathologies, including Alzheimer's disease dementia (ADD), vascular dementia (VaD) and frontotemporal dementia (FTD). People may benefit from knowing the type of dementia they live with, as this could inform prognosis and may allow for tailored treatment. Beta‐amyloid (1‐42) (ABeta42) is a protein which decreases in both the plasma and cerebrospinal fluid (CSF) of people living with ADD, when compared to people with no dementia. However, it is not clear if changes in ABeta42 are specific to ADD or if they are also seen in other types of dementia. It is possible that ABeta42 could help differentiate ADD from other dementia subtypes.

Objectives

To determine the accuracy of plasma and CSF ABeta42 for distinguishing ADD from other dementia subtypes in people who meet the criteria for a dementia syndrome.

Search methods

We searched MEDLINE, and nine other databases up to 18 February 2020. We checked reference lists of any relevant systematic reviews to identify additional studies.

Selection criteria

We considered cross‐sectional studies that differentiated people with ADD from other dementia subtypes. Eligible studies required measurement of participant plasma or CSF ABeta42 levels and clinical assessment for dementia subtype.

Data collection and analysis

Seven review authors working independently screened the titles and abstracts generated by the searches. We collected data on study characteristics and test accuracy. We used the second version of the 'Quality Assessment of Diagnostic Accuracy Studies' (QUADAS‐2) tool to assess internal and external validity of results. We extracted data into 2 x 2 tables, cross‐tabulating index test results (ABeta42) with the reference standard (diagnostic criteria for each dementia subtype). We performed meta‐analyses using bivariate, random‐effects models. We calculated pooled estimates of sensitivity, specificity, positive predictive values, positive and negative likelihood ratios, and corresponding 95% confidence intervals (CIs).

In the primary analysis, we assessed accuracy of plasma or CSF ABeta42 for distinguishing ADD from other mixed dementia types (non‐ADD). We then assessed accuracy of ABeta42 for differentiating ADD from specific dementia types: VaD, FTD, dementia with Lewy bodies (DLB), alcohol‐related cognitive disorder (ARCD), Creutzfeldt‐Jakob disease (CJD) and normal pressure hydrocephalus (NPH). To determine test‐positive cases, we used the ABeta42 thresholds employed in the respective primary studies. We then performed sensitivity analyses restricted to those studies that used common thresholds for ABeta42.

Main results

We identified 39 studies (5000 participants) that used CSF ABeta42 levels to differentiate ADD from other subtypes of dementia. No studies of plasma ABeta42 met the inclusion criteria. No studies were rated as low risk of bias across all QUADAS‐2 domains. High risk of bias was found predominantly in the domains of patient selection (28 studies) and index test (25 studies).

The pooled estimates for differentiating ADD from other dementia subtypes were as follows: ADD from non‐ADD: sensitivity 79% (95% CI 0.73 to 0.85), specificity 60% (95% CI 0.52 to 0.67), 13 studies, 1704 participants, 880 participants with ADD; ADD from VaD: sensitivity 79% (95% CI 0.75 to 0.83), specificity 69% (95% CI 0.55 to 0.81), 11 studies, 1151 participants, 941 participants with ADD; ADD from FTD: sensitivity 85% (95% CI 0.79 to 0.89), specificity 72% (95% CI 0.55 to 0.84), 17 studies, 1948 participants, 1371 participants with ADD; ADD from DLB: sensitivity 76% (95% CI 0.69 to 0.82), specificity 67% (95% CI 0.52 to 0.79), nine studies, 1929 participants, 1521 participants with ADD. Across all dementia subtypes, sensitivity was greater than specificity, and the balance of sensitivity and specificity was dependent on the threshold used to define test positivity.

Authors' conclusions

Our review indicates that measuring ABeta42 levels in CSF may help differentiate ADD from other dementia subtypes, but the test is imperfect and tends to misdiagnose those with non‐ADD as having ADD. We would caution against the use of CSF ABeta42 alone for dementia classification. However, ABeta42 may have value as an adjunct to a full clinical assessment, to aid dementia diagnosis.

Plain language summary

How accurate is the ABeta42 test for distinguishing Alzheimer's disease from other types of dementia in patients seen in a specialist clinic?

Why is improving dementia diagnosis important?

Dementia is a condition characterised by progressive problems with memory and thinking. Dementia can be caused be a number of different conditions (for example, by Alzheimer's disease), and the best treatments depend on the underlying cause. Levels of the protein ABeta42 in blood or spinal fluid may determine the underlying cause of dementia. This could help clinicians choose the best treatments.

What is the aim of this review?

The aim of this review was to find out how accurate are the levels of ABeta42 in blood or spinal fluid for determining the cause of dementia.

What was studied in the review?

We included studies that examined the levels of ABeta42 taken from samples of blood or spinal fluid. At present, this test is only used in specialist clinics. Levels of ABeta42 may be lower in persons with Alzheimer's dementia compared to those with other types of dementia.

What are the main results of this review?

We included 39 studies with a total of 5000 participants. All studies used spinal fluid tests of ABeta42. None of the included studies used a blood test of ABeta42.

In theory, the results of these studies indicate that if ABeta42 were to be used in a specialist clinic in a group of 1000 people, where 520 (52%) have Alzheimer's dementia, an estimated 602 would have an ABeta42 result. This would indicate that Alzheimer's dementia is present. Of these, 192 (32%) would be incorrectly classified as having Alzheimer's disease. Of the 398 people with a result indicating that Alzheimer's disease is not present, 110 (28%) would be incorrectly classified as not having Alzheimer's disease. The included studies used different levels of ABeta42 to make the diagnosis of Alzheimer's disease, and the accuracy of the test depended on the level of ABeta42 used.

How reliable are the results of the studies in this review?

In most of the included studies, the diagnosis of Alzheimer's dementia was made by assessing all participants with standard diagnostic criteria.This is likely to have been a reliable method for deciding whether patients really had Alzheimer's disease. However, there were some problems with how the studies were conducted. This may result in ABeta42 appearing more accurate than it really is.

To whom do the results of this review apply?

The results apply to patients undergoing dementia assessment in a specialist setting.

What are the implications of this review?

Measuring levels of ABeta42 in spinal fluid may help distinguish Alzheimer’s disease from other types of dementia, but the test is not perfect. ABeta42 is unlikely to be used in isolation for making a diagnosis, and may have greatest value when used in addition to the other assessments and tests that are undertaken to make a diagnosis of dementia.

How up‐to‐date is the review?

The review authors searched for and included studies published up to February 2020.

Summary of findings

Background

Dementia is a syndrome of chronic decline in cognitive abilities severe enough to impair function in everyday activities (Robinson 2015). The ageing population will lead to an increased prevalence of neurodegenerative diseases such as dementia, with substantial implications for economies and society. Dementia has an annual estimated cost of over USD 818 billion worldwide (Prince 2015).

Dementia is a clinical syndrome that may have multiple aetiologies (DeTure 2019). Alzheimer's disease dementia (ADD) is the most common dementia subtype, affecting 6% of individuals over the age of 65 and 20% over the age of 80 (Knapp 2007). In terms of prevalence, it is followed by vascular dementia (VaD), mixed ADD/VaD, dementia with Lewy bodies (DLB), alcohol‐related dementia and frontotemporal dementia (FTD) (Lopes 2010). In practice and in research, it can be difficult to differentiate between dementia subtypes (Karantzoulis 2011; Ryan 2018). There is often considerable overlap in the presentation with many common clinical features across the dementia subtypes (Karantzoulis 2011). Clinical diagnosis of dementia subtype is imperfect and diagnosis of ADD and other related disorders based on clinical criteria alone does not always align with the diagnosis made on neuropathology at autopsy (Beach 2012). However, differentiating subtypes is important for clinical practice. A pathological diagnosis of dementia type can guide personalised treatments and inform discussions around prognosis (Karantzoulis 2011). Medications approved for symptomatic treatment of dementia, such as cholinesterase inhibitors, are only recommended in certain dementia types. It is also possible that new treatments under development may have differential efficacy across dementia types (Karantzoulis 2011).

In Alzheimer's disease, amyloid beta peptides (ABeta) are produced via sequential cleavage, involving the action of beta and gamma secretases (De Strooper 2010). The most prevalent ABeta species produced during amyloid precursor protein processing are ABeta40 and ABeta42 (Murphy 2010). Amyloid deposition in the brain is a hallmark of Alzheimer's disease. The amyloid hypothesis of Alzheimer's disease describes a pathological cascade process resulting in the aggregation of soluble ABeta42 into insoluble oligomers and then plaques (Takami 2009). Measuring ABeta has been proposed as a diagnostic biomarker, as these proteins may reflect the underlying pathology of Alzheimer's disease (Hansson 2019). ABeta42 in cerebrospinal fluid (CSF) is a biomarker that is entering research and practice, and is said to reflect amyloid plaque burden in the brain (Hansson 2019). There is increasing evidence to suggest that the neurobiology underlying ADD is associated with reductions in ABeta42 levels in CSF (Hansson 2019). Although CSF ABeta42 reductions have been clearly associated with ADD, it is not yet clear if these changes are specific to ADD, or are a marker of other neurodegenerative processes (Hansson 2019). While most amyloid beta research has used CSF, it has been recently demonstrated that plasma markers of ABeta42 may have utility (Nakamura 2018).

Use of ABeta42 is increasing in clinical research of agents that target specific components of the amyloid neuropathological cascade. However, the association between ABeta42 levels and clinical dementia is not fully understood. People can have substantial cortical amyloid without developing clinical symptoms (Jansen WJ 2015) and individuals display variation in their resilience to the presence of cortical amyloid. Amyloid beta itself may not be the pathological entity and amyloidosis triggers downstream pathological processes that drive neurodegeneration and neuronal dysfunction, e.g. tau aggregation (Blurton‐Jones 2006). It has also been postulated that amyloidosis may need the co‐occurrence of another insult, e.g. cerebrovascular disease, to mediate clinical symptomatology (DeTure 2019; Klohs 2019).

While previous Cochrane reviews have sought to understand the value of abnormal levels of cortical amyloid to predict decline from a prodromal to a dementia phase of Alzheimer’s disease (Ritchie 2014; Ritchie 2017), this review focussed on the ability of ABeta42 measures to differentiate between ADD from other dementia types.

Target condition being diagnosed

In this review we considered ADD and other pathological subtypes of dementia. We considered non‐ADD subtypes as a group, and then considered separate pathological diagnoses within that group.

1) ADD

Alzheimer's disease is thought to underlie ADD. Alzheimer's disease is a clinical syndrome that manifests as progressive memory decline, with impairment in at least one other domain of cognitive function, which impacts on the person's function and behaviour (Karantzoulis 2011; Ryan 2018). Alzheimer's pathology affects the limbic system (primarily the hippocampus) and other mesiotemporal structures (DeTure 2019). The pathology also extends to other regions of the neocortex, including the frontal and parietal lobes, generating executive dysfunction and problems with praxis respectively (DeTure 2019; Karantzoulis 2011). Over time, the patient will develop worsening functional impairment as a consequence of their cognitive symptoms (Wilkosz 2010). Criteria such as those of the National Institute of Neurological and Communicative Diseases and Stroke, and the Alzheimer's Disease and Related Disorders Association (the NINCDS‐ADRDA Alzheimer's Criteria 1984) and the Diagnostic and Statistical Manual of Mental Disorders (DSM) are currently used for the differential diagnosis of other dementia subtypes from ADD (Dubois 2007) (Appendix 1).

2) VaD

VaD is caused by underlying cerebrovascular disease (Burns 2005). Vascular dementia tends to follow a stepwise deterioration that is unpredictable in both speed of progression and clinical features (Iadecola 2019). The diagnosis for probable vascular dementia is based on criteria such as those of National Institute of Neurological Disorders and Stroke and the Association Internationale pour la Recherché et l'Enseignement en Neurosciences (the NINDS‐AIREN criteria) (Roman 1993). These criteria have 58% sensitivity and 80% specificity for differentiating VaD from other dementias (Appendix 1).

3) FTD

FTD is the second most common form of dementia in people below the age of 65 years. FTD is associated with progressive change in personality, behaviour and language (Young 2018). Frontotemporal dementias tend to affect planning, judgement, personality and language early (Karantzoulis 2011; Young 2018). Memory impairment is not a prominent feature but by late stage, multiple cognitive domains may be affected (Karantzoulis 2011; Young 2018). The mean sensitivity and specificity for the Lund and Manchester criteria for differentiating FTD from other dementia subtypes were both 97% (Lopez 1999) (Appendix 1). Within the FTD classification, there are subgroups of disease with differing risk factors, pathology and presentation.

4) DLB

In DLB, the characteristic pathology responsible for neurodegeneration in vulnerable neuronal populations is the presence of alpha‐synuclein and ubiquitin aggregates within intraneuronal inclusion bodies, known as Lewy bodies (Outeiro 2019). These consist of a dense granular core, surrounded by a halo of radiating filaments (Beyer 2009). DLB principally leads to impairment in attention, with prominent, early neuropsychiatric symptoms (Outeiro 2019). According to Braak's and McKeith's staging/categorisation systems, the pathology correlates with clinical symptoms such that brainstem pathology is responsible for the extrapyramidal effects, whereas dementia results from neocortical pathology (Parkkinen 2008). The sensitivity and specificity of McKeith's 1996 clinical diagnostic criteria for differentiating DLB from other dementias was 60% and 94% respectively, while McKeith's 2005 criteria give sensitivity and specificity of 91% and 67% respectively (Rizzo 2018). Thus, clinical diagnostic criteria have become more sensitive and less specific over time (Appendix 1).

5) Dementia caused by alcohol‐related cognitive disorder (ARCD)

Dementia originating primarily from chronic alcohol abuse or secondarily by alcohol‐related syndromes, such as Wernicke's encephalopathy, is a common form of dementia in older individuals (Thomas 2001). The similarities between ADD and ethanol‐related neurodegeneration, in addition to the higher prevalence of ADD in older patients, and the reluctance to admit alcohol excess, makes differentiating the two problematic (Kril 1999). The clinical diagnosis of 'alcohol induced persisting dementia' (Kapaki 2005) is based on the criteria set out in the DSM, 4th edition (DSM‐IV) (APA 2000) (Appendix 1).

6) Dementia caused by CJD

Sporadic CJD and Alzheimer's disease share some clinical features, although the former is characterised by rapidly progressive dementia (Otto 2000). The International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD‐10) (WHO 1993) clinical criteria, such as clinical symptoms and characteristic electroencephalography (EEG), are used for diagnosis of CJD, including the presence of 14‐3‐3 protein in CSF, with 84% sensitivity and 92% specificity (Van Everbroeck 1999) (Appendix 1).

7) Dementia caused by NPH

NPH is classically characterized by the triad of symptoms, namely gait disturbance, dementia and urinary incontinence, and is associated with brain ventricular enlargement (Hakim 1965). NPH is one of the few known treatable causes of dementia. Thus, the discrimination of patients with dementia caused by NPH from patients with ADD or VaD is important, as dementia in early stage NPH is considered surgically reversible (Kapaki 2007).

Index test(s)

Our index test is a quantitative measure of ABeta42, measured in either CSF or blood. The assays commonly used to measure ABeta42 levels are the Innogenetics INNOTEST beta‐amyloid 1‐42 kit and the Athena Diagnostics test.

Clinical pathway

Dementia symptoms can develop slowly and only become obvious when there is marked cognitive impairment. Early assessment of cognitive issues would usually be in primary care or a generalist setting, with referral to a specialist dementia service as needed. The differentiation of dementia subtype would usually be performed in specialist, secondary care services. If CSF samples were to be used, this would necessarily be the reserve of the specialist clinic (NICE 2018), due to the invasive nature of these samples. Thus, our question relates to later stages in the clinical pathway, when people are already diagnosed with suspected, but undifferentiated, dementia. The potential use of the ABeta42 biomarkers that we consider in this review would be to differentiate dementia subtype, allowing individualised treatment (Khoury 2019).

ABeta42 testing is not standard practice in clinical settings. Using measures of amyloid in people with suspected neurodegenerative disease has been the subject of a substantial amount of research (Fantoni 2018; Ossenkoppele 2015; Ritchie 2014; Ritchie 2017) and debate within the dementia community. To date, the low specificity of abnormal ABeta42 levels in CSF has limited the clinical uptake of this biomarker (O'Brien 2017). The situation is different in research, and use of amyloid beta biomarkers to identify participants for anti‐amyloid therapies is now obligatory in certain disease‐modifying ADD trials (Cummings 2019). Even in this context, ABeta42 in isolation is imperfect as a case‐mix adjuster or method for ensuring a pure ADD population (Hansson 2019; Niemantsverdriet 2017; Ritchie 2014).

Alternative test(s)

There are other methods for quantifying amyloid burden in the brain, e.g. neuroimaging using positron emission tomography (PET). For the purposes of this review, we focused only on CSF or blood testing of ABeta42 (Rabinovici 2019).

Rationale

Research criteria for defining the pathological process of Alzheimer's disease incorporate and promote use of biomarkers that can quantify amyloid burden. In clinical trials, ABeta42 is used to select potential participants. The use of CSF biomarkers, while not routine, is increasing in clinical practice (Albert 2011; Dubois 2010; McKhann 2011). However, before we incorporate biomarkers into practice or research it is crucial that we understand their diagnostic accuracy.

In this review, we considered ABeta42 as a tool for differentiating dementia subtypes. If a test could classify people with dementia based on the underlying pathology, this could have utility in clinical practice. It would allow tailored treatment (for example cholinesterase inhibitors work well in ADD but less well in VaD) and could be used to inform discussions around prognosis. A tool to classify dementia subtype would also have utility in research. Treatments are being developed that are specific to certain pathological processes, and tools such as ABeta42 could help ensure that the participants enrolled in trials are those with the pathology most likely to benefit from the intervention.

Objectives

To determine the diagnostic accuracy of plasma and CSF ABeta42 for distinguishing ADD from other forms of dementia in people who meet the general diagnostic criteria for a dementia syndrome in a specialist care setting

Secondary objectives

To determine the diagnostic accuracy of plasma and CSF ABeta42 for distinguishing Alzheimer's disease dementia from specific forms of dementia (VaD, FTD, DLB, ARCD, CJD, NPH) in people who meet the general diagnostic criteria for a dementia syndrome in a specialist care setting.
To investigate the effect of ABeta42 thresholds used to define test positivity on the test accuracy reported

Methods

Criteria for considering studies for this review

Types of studies

We considered cross‐sectional studies and noted the timeframe between the clinical diagnostic criteria and the ABeta42 measurement. In line with our review question, we only considered studies in which people with ADD were differentiated from patients with other dementia subtypes and not from cognitively healthy controls. In some studies, the final diagnosis was only confirmed after one to two years of follow‐up, where CSF samples taken at the initial assessment were retrospectively analyzed. We considered these delayed verification studies eligible for inclusion in the review. We limited our inclusion to English‐language studies.

Participants

We included all participants with a clinical diagnosis of any form of dementia, made using the standard clinical diagnostic criteria (Appendix 1) for the respective dementia subtype. We did not include participants with mild cognitive impairment. The setting of interest was specialist dementia services, whether serving outpatients or inpatients.

Index tests

Our index test is a quantitative measure of ABeta42, measured in either CSF or blood. There is currently no consensus on the threshold value that should signify test positivity for plasma or CSF ABeta42 tests. For our analyses, we did not pre‐specify the positivity threshold, but used the thresholds that informed the primary analyses in the respective individual studies. We classified participants assessed by ABeta42 biomarkers as either test‐positive (below study‐specific threshold) or test‐negative (above study‐specific threshold) at baseline. We accepted any assay used to quantify the ABeta42.

Target conditions

Target conditions in this review are as follows.

ADD and non‐ADD, considered in aggregate and then considered by specific diagnoses:
- VaD
- FTD
- DLB
- Dementia caused by ARCD
- Dementia caused by CJD
- Dementia caused by NPH

Reference standards

For the purpose of this review, we accepted any validated clinical criteria‐based definition of dementia, including iterations of DSM and ICD (APA 1987; APA 1994; WHO 1993) (Appendix 1). For ADD, we also accepted the NINCDS‐ADRDA criteria (McKhann 1984).

Diagnostic criteria used to establish the other dementia subtypes in those participants with non‐ADD were as follows:

for VaD: the NINDS‐ARIEN criteria (Roman 1993), the Alzheimer's Disease Diagnostic and Treatment Centers (ADDTC) criteria (Chui 1992), DSM‐III‐R criteria, DSM‐IV criteria or ICD criteria;
for FTD: the Lund criteria (Lund Manchester Groups 1994), Neary 1998 criteria or Boxer 2005 criteria;
for DLB: the reference standard is the McKeith criteria (McKeith 1996,McKeith 2002 or McKeith 2005);
for ARCD: the diagnostic criteria should follow DSM‐III‐R or DSM‐IV;
for dementia in CJD: the ICD‐10 clinical criteria and characteristic EEG should be used;
for dementia caused by NPH: we accepted ICD or DSM criteria.

Search methods for identification of studies

We used a variety of information sources to ensure all relevant studies are included. The Information Specialist of the Cochrane Dementia and Cognitive Improvement Group devised the search strategies for electronic database searching.

Electronic searches

The most recent searches for this review were performed on 18 February 2020. We searched the following databases.

MEDLINE (OvidSP); earliest records to 18 February 2020
Embase (OvidSP); earliest records to 18 February 2020
BIOSIS Previews (Thomson Reuters Web of Science); earliest records to 18 February 2020
Web of Science Core Collection, including Conference Proceedings Citation Index (Thomson Reuters Web of Science); earliest records to 18 February 2020
PsycINFO (OvidSP); earliest records to 18 February 2020
LILACS (Latin American and Caribbean Health Science Information database); earliest records to 18 February 2020

See Appendix 2 for details of the sources searched, the search strategies used, and the number of records that were retrieved.

We did not apply any language or date restrictions to the electronic searches. We did not use methodological search filters (i.e. collections of terms aimed at reducing the number needed to screen by filtering out irrelevant records and retaining only those that are relevant) that were designed to retrieve diagnostic test accuracy studies, because available filters have not yet proved sensitive enough for systematic review searches (Beynon 2013).

Searching other resources

We also conducted searches in the following databases for other related systematic diagnostic accuracy reviews.

Meta‐analyses van Diagnostisch Onderzoek (MEDION) (www.mediondatabase.nl)
Database of Abstracts of Reviews of Effects (DARE) (www.york.ac.uk/inst/crd/crddatabases.htm#DARE),
Health Technology Assessments Database (HTA Database) (www.york.ac.uk/inst/crd/crddatabases.htm#HTA)
Aggressive Research Intelligence Facility (ARIF) database (www.arif.bham.ac.uk)

We searched for systematic reviews of diagnostic studies from the International Federation of Clinical Chemistry and Laboratory Medicine Committee for Evidence‐based Laboratory Medicine database (C‐EBLM). We checked reference lists of any relevant systematic reviews for additional studies.

Data collection and analysis

Selection of studies

One review author (ANS) screened all titles and abstracts generated by electronic database searches for relevance, and excluded duplicate records. Following de‐duplication, second assessment of the search results was divided among seven review authors (MK, RW, AH, MD, AG, EP, AA, LB, and TQ). Pairs of review authors (from among LB, MK, NS, and TQ) independently assessed full manuscripts against the inclusion criteria. Where necessary, a third review author (CR) resolved disagreements. The search was updated on 18 February 2020. When the same dataset was presented in more than one paper, we included the primary paper, which was the paper with the largest number of patients or with the most informative data.

Data extraction and management

We extracted data on study characteristics into a pre‐standardised data extraction form, including data for the assessment of study quality and data for investigation of heterogeneity, as described in Appendix 3. We also extracted data for creating 2 x 2 tables (cross‐relating index test results to the reference standards). Data extraction was performed independently by four blinded review authors (MK, NS, LB, TQ). Disagreement in data extraction was resolved by discussion, involving a third review author (CR) as arbitrator when necessary. Where a study did not present all relevant data for creating a 2 x 2 table, we contacted the study authors directly to request further information.

Assessment of methodological quality

We assessed the methodological quality of each study using the second version of the 'Quality Assessment of Diagnostic Accuracy Studies' (QUADAS‐2) tool (Whiting 2011). The tool is made up of four domains: patient selection; index test; reference standard; and patient flow. Four independent raters (MK, NS, LB, TQ), blinded to each other’s scores, performed QUADAS‐2 assessments. Disagreement was resolved by further review and discussion with potential to involve a third review author (CR) as arbitrator if necessary. We assessed each domain in terms of risk of bias, with the first three domains also considered in terms of applicability. The components of each of these domains, and a rubric that details how judgements concerning risk of bias are made, are detailed in Appendix 4 and Appendix 5. We produced a narrative summary, describing numbers of studies that were found to have high, low, or unclear risk of bias, as well as describing our concerns regarding applicability.

Statistical analysis and data synthesis

We extracted the data from each study into a 2 x 2 table, showing the binary test results cross‐classified with the binary reference standard. We organised test data so that the reference standard was always ADD and thus accuracy data were around differentiating ADD from other dementias. We entered true positive (TP), false negative (FN), false positive (FP) and true negative (TN) data from the included studies into RevMan 5.4 (Cochrane 2020) to calculate sensitivity and specificity and their 95% confidence intervals. We performed summary analyses using bivariate random‐effects models, based on pairs of sensitivity and specificity, to calculate pooled estimates of sensitivity, specificity, positive predictive values, positive likelihood ratios and negative likelihood ratios, all with their associated 95% confidence intervals.

We used version 1.2 of the MetaDTA diagnostic test accuracy meta‐analytic software (Freeman 2019; Patel 2020) in our analyses.

We presented summary analyses as forest plots and in receiver operating characteristic (ROC) space by plotting estimates of sensitivity and specificity with the associated 95% confidence interval of the pooled estimate. We only performed meta‐analyses where there were sufficient studies (three or more studies).

Investigations of heterogeneity

We described the following factors:

Index test: i) thresholds used; ii) method used to measure ABeta42 levels;
Target disorder: i) reference standard used, e.g. NINCDS‐ADRDA criteria versus DSM criteria versus ICD‐10 criteria for ADD; ii) criteria used for the definition of a dementia syndrome: e.g. individual, clinician, algorithm, or consensus group
Target population: i) spectrum of patients: age, sex, education, sampling strategy, Mini‐Mental State Examination (MMSE) score and Apolipoprotein E (APOE) status of study participants; ii) clinical setting: outpatients versus inpatients versus participants in residential care.

Sensitivity analyses

We performed sensitivity analyses to assess the effect of differing ABeta42 test thresholds. In comparisons of ADD versus non‐ADD, ADD versus VaD, ADD versus FTD, and ADD versus DLB, we grouped studies by similar thresholds as follows: those using thresholds less than or equal to 500 pg/ml, and those using thresholds over 500 pg/ml. We performed sensitivity analyses only where there were sufficient studies (three or more studies) to do so.

In addition, we performed sensitivity analyses for studies with younger populations of ADD participants: those where the mean age was under 66 years or who specifically enrolled participants with early‐onset ADD.

We performed subgroup analyses on FTD variants: behavioural variant (bvFTD), and primary progressive aphasia (PPA).

Finally, we performed sensitivity analyses for studies with high drop out rates (greater than 30% of participants), and those which not pre‐specify the test threshold.

Assessment of reporting bias

We did not investigate reporting bias because of current uncertainty about how it operates in test accuracy studies, and concerns about the interpretation of existing analytical tools, such as funnel plots.

Results

Results of the search

We identified 57,763 titles after the electronic searches (Figure 1). After de‐duplication and screening of titles for relevance, we screened 34,027 abstracts. We assessed 1835 full papers for eligibility and included 39 papers in the review.

Study flow diagram through the screening process.

We contacted seven authors for additional information about their studies but did not obtain usable data (Brandt 2008; Carandini 2019; Hampel 2018; Smach 2008a; Toledo 2012; van Steenoven 2018; van Steenoven 2019).

Summary of included studies

The Characteristics of included studies table lists the characteristics of the 39 included studies, comprising a total of 7246 participants. All studies were published between 2000 and 2020. Thirty‐five studies were conducted in Europe. Three studies (Montine 2001; Shi 2018; Tariciotti 2018) were conducted in the USA and one study (Smach 2008) was conducted in Tunisia.

Index test

For the method used to measure ABeta42 levels (Table 3), 31 studies used the Innogenetics ELISA kit. Two studies used INNOTEST β‐AMYLOID (1‐42) ELISA kits from Fujirebio Inc. (Casoli 2019; Marchegiani 2019). One study used Athena Diagnostics (Montine 2001), one study used the ADmark ELISA kit (Tariciotti 2018) and one study used the ABeta‐SDS‐Page Immunoblot (Wiltfang 2003). Two studies did not report the ELISA kit they used (Lombardi 2018; Schirinzi 2015).

1. Included studies and the index test accuracy at study level.

Included studies and the accuracy of CSF Aβ42 for discriminating ADD from other dementia subtypes
Differential diagnosis	Study	Participants N (included in analysis)	Threshold assays	Threshold pre‐specified	Test accuracy at study level
Differential diagnosis	Study	Participants N (included in analysis)	Threshold assays	Threshold pre‐specified	Sensitivity (%)	Specificity (%)
ADD versus non‐ADD	Brettschneider 2006	N = 165: 109 ADD; 56 non‐ADD (41 VaD; 15 FTD)	612 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	82%	46%
	Kapaki 2003	N = 64: 49 ADD; 15 non‐ADD (6 DLB; 4 FTD; 1 PDD; 2 PSP; 2 CBGD)	435 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	71%	80%
	Knapskog 2018	N = 155: 138 ADD; 17 non‐ADD (subtypes not specified)	550 pg/ml and 700 pg/ml ELISA, Innogenetics, Ghent, Belgium	Yes	43% and 35%	79% and 47%
	Lewczuk 2004	N = 33: 21 ADD; 11 non‐ADD (5 VaD; 1 mixed; 1 SCASE; 1 SD; 1 FTD; 1 ARCD; 1 unspecified)	500 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	86%	82%
	Lombardi 2018	N = 45: 32 ADD; 10 FTD; 3 unclassified cognitive decline	650 pg/ml 600 pg/ml ELISA (unspecified)	Yes No	73% and 87%	64%
	Maddalena 2003	N = 81: 51 ADD; 30 non‐ADD (8 VaD; 3 FTD; 2 DLB; 2 PDD; 2 CJD; 2 CAA; 11 other)	490 pg/ml ELISA, Innogenetics, Belgium	No	78%	70%
	Montine 2001	N = 27: 19 ADD; 8 non‐ADD (1 DLB; 3 NPH; 3 PPA; 1 hippocampal sclerosis)	1125 pg/ml Athena Diagnostics, Worcester, MA, USA.	Yes	100%	25%
	Rosler 2001	N = 51: 27 ADD (11 EO; 16LO); 24 non‐AD (5 VaD; 4 PDD; 2 LBD; 8 NPH; 5 other)	375 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	78%	58%
	Smach 2008	N = 108: 73 ADD; 35 non‐ADD (18 VaD; 5 FTD; 3 DLB; 7 mixed; 2 unclassified)	505 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	82%	71%
	Spies 2010	N = 138: 69 ADD; 69 non‐ADD (26 VaD; 27 FD; 16 DLB)	Threshold not reported ELISA, Innogenetics NV, Ghent, Belgium	No	83%	74%
	Tapiola 2000	N = 107: 80 probable ADD; 27 non‐ADD (8 VaD; 4 FTD; 5 LBD; 3 PDD; 7 unclassified) Note: 41 definite ADD not included in analysis	340 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	69%	59%
	Tariciotti 2018	N = 749: 264 ADD; 485 non‐ADD (65 DLB, 53 FTD, 31 VaD, 21 PSP, 14 CBD, 218 NPH, 30 CJD) Note: 121 uncertain diagnosis not included in analysis	500 pg/ml ADmark ELISA kit	Yes	81%	54%
	Perani 2016	N = 75: 47 ADD; 28 non‐ADD (14 FTLD; 14 DLB)	500 ng/L ELISA, Innogenetics, Ghent, Belgium	Yes	85%	46%
ADD versus VaD	De Jong 2006	N = 86: 61 ADD; 25 VaD	520 pg/ml Innogenetics NV, Ghent, Belgium	No	82%	76%
	Kapaki 2003	N = 55: 49 ADD; 6 VaD	526 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	82%	67%
	Lins 2004	N = 24: 12 ADD; 12 VaD	562 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	67%	50%
	Marchegiani 2019	N = 87: 70 ADD, 17 VaD	431 pg/ml ELISA, Fujirebio Inc., Tokyo, Japan	No	65%	95%
	Paraskevas 2009	N = 115: 92 ADD; 23 VaD	461 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	78%	70%
	Sjogren 2000	N = 85: 60 ADD (37 EO; 23 LO); 24 VaD (SWM dementia)	537 pg/ml ELISA, Innotest, Innogenetics, Belgium	Yes	93%	33%
	Spies 2010	N = 95: 69 ADD; 26 VaD	Threshold not reported ELISA, Innogenetics NV, Ghent, Belgium	No	83%	69%
	Stefani 2005	N = 55: 35 ADD; 20 VaD	493 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	77%	80%
	Herbert 2014	N = 79: 64 ADD; 15 VaD	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	70%	87%
	Tariciotti 2018	N = 295: 264 ADD; 31 VaD (Note: 121 uncertain diagnosis not included in analysis)	500 pg/ml ADmark ELISA kit	Yes	81%	39%
	Santangelo 2017	N = 176: 165 ADD; 11 VaD	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	82%	82%
ADD versus FTD	de Rino 2012	N = 114: 72 ADD; 42 bvFTD	104 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	82%	21%
	Abu‐Rumeileh 2018	N = 113: 60 ADD; 53 bvFTD (Note: 10 FTD not included in analysis)	482 pg/ml Innotest, Innogenetics, Ghent, Belgium	No	89%	80%
	Bibl 2007	N = 60: 30 ADD; 30 FTD (FTLD: 24 FTD; 5 PPA; 1 SD)	Threshold not reported ELISA	No	90%	90%
	Casoli 2019	N = 76: 55 ADD; 21 FTD (12 bvFTD and 9 PPA)	Various (minimum threshold 112 maximum 1006 and 837 pg/ml) ELISA, Fujirebio Inc., Tokyo, Japan	No	100%	0%
	Falgas 2020	N = 90: 64 AD; 26 FTD Note: only 23 (18 FTD and 5 ADD) included in the analysis as MCI excluded	494.95 pg/ml Innotest, Innogenetics, Ghent, Belgium	No	100%	94%
	Kapaki 2008	N = 107: 76 ADD; 31 FTD (FTLD: 24 FTD; 7 PPA & FTD) Note: 3 FTLD not included in analysis	≤ 451 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	75%	71%
	Khoonsari 2019	N = 87: 76 ADD; 11 FTD (subtype unspecified)	530 pg/ml Innotest, Innogenetics, Ghent, Belgium	Yes	88%	91%
	Lombardi 2018	N = 45: 32 ADD; 10 FTD (subtype not specified); 3 non‐ADD	650 pg/ml 600 pg/mlELISA (unspecified)	Yes No	73 and 87%	70%
	Marchegiani 2019	N = 93: 70 ADD; 23 FTD (19 FTD, 3 PSP, 3 CBD)	613 pg/ml ELISA, Fujirebio Inc., Tokyo, Japan	No	96%	57%
	Shi 2018	N = 170: 114 ADD; 56 FTD (48 bvFTD, 8 CBS) Note: samples excluded where haemoglobin was >500 ng/ml	Threshold not reported ELISA, Innogenetics NV, Ghent, Belgium	No	80%	80%
	Sjogren 2000	N = 77: 60 ADD (37 EO; 23 LO); 17 FTD	537 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	92%	59%
	Spies 2010	N = 96: 69 ADD; 27 FTD	Threshold not reported ELISA, Innogenetics NV, Ghent, Belgium	No	94%	85%
	Baldeiras 2015	N = 214: 107 ADD; 107 FTD	≤ 538pg/ml ELISA, Innotest, Innogenetics, Belgium	No	70%	82%
	Herbert 2014	N = 90: 64 ADD; 26 FTD	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	70%	88%
	Santangelo 2017	N = 208: 165 ADD; 43 FTD	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	82%	67%
	Tariciotti 2018	N = 317: 264 ADD; 53 FTD (Note: 121 uncertain diagnosis not included in analysis)	500 pg/ml ADmark ELISA kit	Yes	81%%	40%
	Perani 2016	N = 61: 47 AD; 14 FTD	500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	85%	71%
ADD versus DLB	Aerts 2011	N = 65: 44 ADD; 21 DLB	> 482 pg/ml ELISA, Innogenetics NV, Ghent, Belgium	No	62%	65%
	Bibl 2006	N = 41: 18 ADD; 23 DLB	475 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	50%	96%
	Bousiges 2018	N = 937: 783 ADD; 154 DLB	700 np/ml 606 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes No	71% and 85%	53% and 37%
	Spies 2010	N = 85: 69 ADD; 16 DLB	Threshold not reported. ELISA, Innogenetics NV, Ghent, Belgium	No	65%	75%
	Herbert 2014	N = 78: 64 ADD; 14 DLB	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	70.3%	50%
	Santangelo 2017	N = 187: 165 ADD; 22 DLB	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	82%	41%
	Shi 2018	N = 156: 114 ADD; 42 DLB (Note: samples excluded where haemoglobin was > 500 ng/ml)	Threshold not reported ELISA, Innogenetics NV, Ghent, Belgium	No	89%	74%
	Tariciotti 2018	N = 329: 264 ADD; 65 DLB (10 LBD, 32 PDD)	500 pg/ml ADmark ELISA kit	Yes	81%	60%
	Bousiges 2016	N = 51: 31 ADD; 20 DLB	500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	77%	80%
ADD versus CJD dementia	Kapaki 2001	N = 50: 38 ADD; 12 CJD	445 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	76%	42%
	Tariciotti 2018	N = 294: 264 ADD; 30 CJD	500 pg/ml ADmark ELISA kit	Yes	81%	40%
	Wiltfang 2003	N = 38: 19 ADD; 19 CJD	1900 pg/ml Aβ‐SDS‐PAGE immunobolt	No	100%	58%
ADD versus NPH dementia	Kapaki 2007	N = 85: 67 ADD; 18 NPH	> 268 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	91%	44%
	Lins 2004	N = 24: 12 ADD; 12 NPH	562 pg/ml ELISA, Innogenetics, Ghent, Belgium	No	67%	33%
	Schirinzi 2015	N = 28: 14 ADD; 14 NPH	371 pg/ml ELISA (unspecified)	No	73.3%	81.3%
	Santangelo 2017	N = 199: 165 ADD; 34 NPH	≤ 500pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	Yes	82%	26%
ADD versus ARCD dementia	Kapaki 2005	N = 53: 33 ADD; 20 ACRD	≤ 562 pg/ml ELISA, Innotest, Innogenetics, Ghent, Belgium	No	85%	80%

Patient population	People with a clinical diagnosis of dementia.
Review question	How accurate is CSF ABeta42 test for distinguishing Alzheimer's disease dementia (ADD) from other types of dementia?
Index test	Cerebrospinal fluid (CSF) ABeta42 test.
Reference standard	Clinical diagnostic criteria for dementia pathological subtypes (Appendix 1).
Target condition	ADD vs other dementia subtypes
Included studies	39 studies (5000 participants).
Quality concerns	The majority of studies were classified at high or unclear risk of bias, particularly for patient selection (n = 28), and the index test (n = 25). The majority of studies were at low risk for applicability concerns (n = 33 to 36 for each domain). Studies were mainly at unclear risk of bias due to inadequate reporting. Few studies pre‐specified the test threshold and used optimal cut‐offs calculated using the study data.
Heterogeneity	The majority of studies in this review were conducted in specialist secondary care settings. The majority of studies conducted the index test in a similar manner. Sources of heterogeneity were: patient population and dementia subtype enrolled, test threshold used, and the diagnostic criteria and definition of ADD and dementia subtypes.
Differential Diagnosis	Number of participants	Number of studies	Number of participants with ADD	Pooled sensitivity (95% confidence interval)	Pooled specificity (95% confidence interval)	Pooled false positive rate (95% confidence interval)	Pooled positive likelihood ratio (95% confidence interval)	Pooled negative likelihood ratio (95% confidence interval)
ADD vs non‐ADD	1704	13	880	0.79 (0.73, 0.85)	0.60 (0.52, 0.67)	0.40 (0.33, 0.48)	1.98 (1.58, 2.47)	0.34 (0.24, 0.49)
ADD vs VaD	1151	11	941	0.79 (0.75, 0.83)	0.69 (0.55, 0.81)	0.31 (0.20, 0.45)	2.58 (1.75, 3.81)	0.30 (0.25, 0.36)
ADD vs FTD	1948	17	1371	0.85 (0.79, 0.89)	0.72 (0.55, 0.84)	0.28 (0.16, 0.45)	3.00 (1.81, 5.00)	0.21 (0.16, 0.28)
ADD vs DLB	1929	9	1521	0.77 (0.70, 0.83)	0.66 (0.51, 0.78)	0.34 (0.22, 0.49)	2.27 (1.57, 3.28)	0.35 (0.28, 0.45)
ADD vs NPH	336	4	258	0.84 (0.79, 0.88)	0.42 (0.26, 0.60)	0.58 (0.40, 0.74)	1.45 (1.07, 1.97)	0.38 (0.23, 0.63)
ADD vs CJD	382	3	321	0.82 (0.77, 0.86)	0.46 (0.34, 0.58)	0.54 (0.42, 0.66)	1.51 (1.15, 1.87)	0.40 (0.26, 0.54)
ADD vs ARCD^a	53	1	33	0.80	0.85	‐	‐	‐
Conclusions	Our results suggest that ABeta42 could be useful in improving differential diagnosis of the dementia syndrome, but the test is imperfect. It is unlikely that the ABeta42 biomarker would be used in isolation in clinical practice and ideally it should be used to support the diagnosis alongside full clinical, radiological, and neuropsychological assessment. Our review does not help answer questions around the added value of the test over routine diagnostics.
Implications	The test accuracy demonstrated does lend some support to the concept of using biomarkers to differentiate dementia type for tailored therapy. Clinical trials of anti‐amyloid interventions could consider using quantification of ABeta42 for patient selection. The biomarker does not guarantee an exclusively ADD population but it may help select those people most likely to benefit from the intervention.

Differential diagnosis	Number of participants	Number of studies	Number of participants with ADD	Pooled sensitivity (95% confidence interval)	Pooled specificity (95% confidence interval)	Pooled false positive rate (95% confidence interval)	Pooled positive likelihood ratio (95% confidence interval)	Pooled negative likelihood ratio (95% confidence interval)
Effect of test threshold
ADD vs non‐ADD (threshold ≤ 500 pg/ml)	1160	7	519	0.79 (0.68, 0.86)	0.58 (0.45, 0.70)	0.42 (0.30, 0.55)	1.87 (1.26, 2.77)	0.37 (0.20, 0.67)
ADD vs non‐ADD (threshold > 500 pg/ml)	406	5	292	0.78 (0.70, 0.84)	0.62 (0.50, 0.73)	0.38 (0.27, 0.50)	2.04 (1.54, 2.71)	0.36 (0.27, 0.49)
ADD vs VaD (threshold ≤ 500 pg/ml)	809	7	697	0.79 (0.74, 0.82)	0.68 (0.51, 0.82)	0.32 (0.19, 0.49)	2.47 (1.54, 3.95)	0.31 (0.25, 0.39)
ADD vs VaD (threshold > 500 pg/ml)	194	3	133	0.86 (0.74, 0.93)	0.65 (0.37, 0.85)	0.35 (0.15, 0.63)	2.43 (1.25, 4.74)	0.22 (0.14, 0.36)
ADD vs FTD (threshold ≤ 500 pg/ml)	1033	8	753	0.87 (0.80, 0.92)	0.51 (0.21, 0.80)	0.49 (0.20, 0.79)	1.77 (0.92, 3.41)	0.25 (0.14, 0.44)
ADD vs FTD (threshold >500 pg/ml)	513	5	345	0.81 (0.73, 0.88)	0.84 (0.72, 0.91)	0.16 (0.09, 0.29)	5.02 (2.66, 9.48)	0.22 (0.14, 0.35)
ADD vs bvFTD (all thresholds)	898	8	651	0.85 (0.80, 0.89)	0.68 (0.51, 0.81)	0.32 (0.19, 0.49)	2.68 (1.65, 4.36)	0.22 (0.15, 0.32)
ADD vs PPA (all thresholds)	192	3	171	0.94 (0.50, 1.00)	0.23 (0.00, 0.98)	0.77 (0.03, 1.00)	1.22 (0.45, 3.34)	0.27 (0.03, 2.71)
ADD vs DLB (threshold ≤ 500 pg/ml)	751	6	563	0.79 (0.69, 0.86)	0.68 (0.46, 0.85)	0.32 (0.15, 0.54)	2.49 (1.37, 4.50)	0.31 (0.22, 0.43)
Effect of age
ADD vs non‐ADD (older participants)	1555	10	779	0.80 (0.76, 0.84)	0.62 (0.52, 0.70)	0.39 (0.30, 0.48)	2.08 (1.66, 2.61)	0.32 (0.26, 0.40)
ADD vs non‐ADD (younger participants)	149	3	105	0.71 (0.47, 0.87)	0.51 (0.32, 0.69)	0.49 (0.31, 0.68)	1.44 (0.78, 2.65)	0.58 (0.22, 1.54)
ADD vs VaD (older participants)	1067	9	881	0.80 (0.75, 0.84)	0.68 (0.53, 0.80)	0.32 (0.20, 0.48)	2.49 (1.65, 3.74)	0.30 (0.25, 0.37)
ADD vs FTD (older participants)	1788	14	1220	0.85 (0.79, 0.90)	0.68 (0.47, 0.84)	0.32 (0.16, 0.53)	2.67 (1.52, 4.69)	0.22 (0.16, 0.30)
ADD vs FTD (younger participants)	160	3	95	0.82 (0.69, 0.91)	0.86 (0.76, 0.93)	0.14 (0.07, 0.25)	6.01 (3.24, 11.14)	0.20 (0.11, 0.38)
Effect of studies with high drop‐out rates removed
ADD vs VaD	896	9	712	0.79 (0.74, 0.84)	0.70 (0.53, 0.83)	0.30 (0.17, 0.47)	2.64 (1,65, 4.24)	0.30 (0.24, 0.36)
ADD vs FTD	1480	14	1023	0.81 (0.76, 0.85)	0.75 (0.62, 0.85)	0.25 (0.15, 0.39)	3.24 (2.05, 5.13)	0.25 (0.20, 0.32)
ADD vs DLB	1929	9	1521	0.745 (0.66, 0.83)	0.68 (0.48, 0.83)	0.33 (0.17, 0.53)	2.32 (1.43, 3.76)	0.37 (0.29, 0.46)
ADD vs NPH	137	3	93	0.86 (0.72, 0.94)	0.49 (0.32, 0.67)	0.51 (0.33, 0.68)	1.70 (1.13, 2.57)	0.28 (0.11, 0.73)
Effect of studies without pre‐specified thresholds removed
ADD vs non‐ADD	566	5	366	0.79 (0.73, 0.84)	0.60 (0.49, 0.71)	0.40 (0.29, 0.51)	1.98 (1.50, 2.62)	0.35 (0.26, 0.46)
ADD vs VaD	265	3	175	0.80 (0.73, 0.85)	0.73 (0.61, 0.82)	0.28 (0.18, 0.39)	1.36 (1.01, 1.71)	0.97 (0.44, 1.50)
ADD vs FTD	870	7	615	0.84 (0.71, 0.92)	0.63 (0.21, 0.91)	0.37 (0.09, 0.79)	2.27 (0.79, 6.57)	0.25 (0.90, 2.47)
ADD vs DLB	214	3	129	0.70 (0.62, 0.76)	0.70 (0.54, 0.82)	0.30 (0.18, 0.46)	2.31 (1.43, 3.75)	0.44 (0.32, 0.59)

NINCDS‐ADRDA (McKhann 1984)	I. The criteria for the clinical diagnosis of PROBABLE Alzheimer's disease include: dementia established by clinical examination and documented by the Mini‐Mental Test, Blessed Dementia Scale, or some similar examination, and confirmed by neuropsychological tests: deficits in two or more areas of cognition: progressive worsening of memory and other cognitive functions; no disturbance of consciousness; onset between ages 40 and 90, most often after age 65; absence of systemic disorders or other brain diseases that in and of themselves could account for the progressive deficits in memory and cognition
	II. The diagnosis of PROBABLE Alzheimer's disease is supported by: progressive deterioration of specific cognitive functions such as language (aphasia), motor skills (apraxia), and perception (agnosia); impaired activities of daily living and altered patterns of behavior; family history of similar disorders, particularly if confirmed neuropathologically; laboratory results of: normal lumbar puncture as evaluated by standard techniques; normal pattern or nonspecific changes in EEG such as increased slow‐wave activity; evidence of cerebral atrophy on CT with progression documented by serial observation.
	III. Other clinical features consistent with the diagnosis of PROBABLE Alzheimer's disease, after exclusion of causes of dementia other than Alzheimer's disease, include: plateaus in the course of progression of the illness: associated symptoms of depression, insomnia, incontinence, delusions, illusions, hallucinations, catastrophic verbal, emotional, or physical outbursts, sexual disorders, and weight loss: other neurologic abnormalities in some patients, especially with more advanced disease and including motor signs such as increased muscle tone, myoclonus, or gait disorder; seizures in advanced disease; CT normal for age.
	IV. Features that make the diagnosis of PROBABLE Alzheimer's disease uncertain or unlikely include: sudden, apoplectic onset; focal neurologic findings such as hemiparesis, sensory loss, visual field deficits, and incoordination early in the course of the illness; seizures or gait disturbances at the onset or very early in the course of the illness.
	V. Clinical diagnosis of POSSIBLE Alzheimer's disease: may be made on the basis of the dementia syndrome, in the absence of other neurologic, psychiatric, or systemic disorders sufficient to cause dementia, and in the presence of variations in the onset, in the presentation, or in the clinical course; may be made in the presence of a second systemic or brain disorder sufficient to produce dementia, which is not considered to be thecause of the dementia; should be used in research studies when a single, gradually progressive severe cognitive deficit is identified in the absence of other identifiable cause.
	VI. Criteria for diagnosis of DEFINITE Alzheimer's disease are: the clinical criteria for probable Alzheimer's disease; histopathologic evidence obtained from a biopsy or autopsy.
	VII. Classification of Alzheimer's disease for research purposes should specify features that may differentiate subtypes of the disorder, such as: familial occurrence; onset before age of 65; presence of trisomy‐21; coexistence of other relevant conditions such as Parkinson's disease.
DSM‐IV (APA 1994); DSM‐IV‐TR (APA 2000)	Diagnostic and Statistical Manual of Mental Disorders (DSM)IV‐TR, A. The Development of multiple cognitive deficits manifested by both: memory impairment (impaired ability to learn new information or to recall previously learned information); and one (or more) of the following cognitive disturbances: (aphasia (language disturbance); apraxia (impaired ability to carry out motor activities despite intact motor function); agnosia (failure to recognize or identify objects despite intact sensory function); disturbance in executive functioning (i.e., planning, organizing, sequencing, abstracting). B. The cognitive deficits in criteria A1 and A2 each cause significant impairment in social or occupational functioning and represent a significant decline from a previous level of functioning. C. The course is characterized by gradual onset and continuing cognitive decline. D. The cognitive deficits in Criteria A1 and A2 are not due to any of the following: other central nervous system conditions that cause progressive deficits in memory and cognition (e.g. cerebrovascular disease, Parkinson's disease, Huntington's disease, subdural hematoma, normal‐pressure hydrocephalus, brain tumour); systemic conditions that are known to cause dementia (e.g. hypothyroidism, vitamin B₁₂ or folic acid deficiency, niacin deficiency, hypercalcaemia, neurosyphilis, HIV infection); substance‐induced conditions. E. The deficits do not occur exclusively during the course of a delirium. F. The disturbance is not better accounted for by another Axis I disorder (e.g. major depressive disorder, schizophrenia). Diagnostic Criteria for 294.1x Dementia of the Alzheimer's Type 294.11 Without Behavioral Disturbance: if the cognitive disturbance is accompanied by a clinically significant behavioral disturbance (e.g. wandering, agitation). Specify subtype: With Early Onset: if onset is at age 65 years or below With Late Onset: if onset is after 65 years
ICD‐10 (WHO 1993)	Diagnostic guidelines The primary requirement for diagnosis is evidence of a decline in both memory and thinking which is sufficient to impair personal activities of daily living, as described above. The impairment of memory typically affects the registration, storage, and retrieval of new information, but previously learned and familiar material may also be lost, particularly in the later stages. Dementia is more than dysmnesia: there is also impairment of thinking and of reasoning capacity, and a reduction in the flow of ideas. The processing of incoming information is impaired, in that the individual finds it increasingly difficult to attend to more than one stimulus at a time, such as taking part in a conversation with several persons, and to shift the focus of attention from one topic to another. If dementia is the sole diagnosis, evidence of clear consciousness is required. However, a double diagnosis of delirium superimposed upon dementia is common (F05.1). The above symptoms and impairments should have been evident forat least 6 months for a confident clinical diagnosis of dementia to be made. Differential diagnosis Consider: a depressive disorder (F30‐F39), which may exhibit many of the features of an early dementia, especially memory impairment, slowed thinking, and lack of spontaneity; delirium (F05); mild or moderate mental retardation (F70‐F71); states of subnormal cognitive functioning attributable to a severely impoverished social environment and limited education; iatrogenic mental disorders due to medication (F06.‐). Dementia may follow any other organic mental disorder classified in this block, or coexist with some of them, notably delirium (see F05.1). F00 Dementia in Alzheimer's disease Alzheimer's disease is a primary degenerative cerebral disease of unknown etiology, with characteristic neuropathological and neurochemical features. It is usually insidious in onset and develops slowly but steadily over a period of years. This period can be as short as 2 or 3 years, but can occasionally be considerably longer. The onset can be in middle adult life or even earlier (Alzheimer's disease with early onset), but the incidence is higher in later life (Alzheimer's disease with late onset). In cases with onset before the age of 65‐70, there is the likelihood of a family history of a similar dementia, a more rapid course, and prominence of features of temporal and parietal lobe damage, including dysphasia or dyspraxia. In cases with a later onset, the course tends to be slower and to be characterized by more general impairment of higher cortical functions. Patients with Down's syndrome are at high risk of developing Alzheimer's disease. There are characteristic changes in the brain: a marked reduction in the population of neurons, particularly in the hippocampus, substantia innominata, locus ceruleus, and temporoparietal and frontal cortex; appearance of neurofibrillary tangles made of paired helical filaments; neuritic (argentophil) plaques, which consist largely of amyloid and show a definite progression in their development (although plaques without amyloid are also known to exist); and granulovacuolar bodies. Neurochemical changes have also been found, including a marked reduction in the enzyme choline acetyltransferase, in acetylcholine itself, and in other neurotransmitters and neuromodulators. As originally described, the clinical features are accompanied by the above brain changes. However it now appears that the two do not always progress in parallel: one may be indisputably present with only minimal evidence of the other. Nevertheless, the clinical features of Alzheimer's disease are such that it is often possible to make a presumptive diagnosis on clinical grounds alone. Dementia in Alzheimer's disease is at present irreversible. Diagnostic guidelines The following features are essential for a definite diagnosis: (a) Presence of a dementia as described above. (b)Insidious onset with slow deterioration. While the onset usually seems difficult to pinpoint in time, realization by others that the defects exist may come suddenly. An apparent plateau may occur in the progression. (c)Absence of clinical evidence, or findings from special investigations, to suggest that the mental state may be due to other systemic or brain disease which can induce a dementia (e.g. hypothyroidism, hypercalcaemia, vitamin B12 deficiency, niacin deficiency, neurosyphilis, normal pressure hydrocephalus, or subdural haematoma). (d)Absence of a sudden, apoplectic onset, or of neurological signs of focal damage such as hemiparesis, sensory loss, visual field defects, and incoordination occurring early in the illness (although these phenomena may be superimposed later). In a certain proportion of cases, the features of Alzheimer's disease and vascular dementia may both be present. In such cases, double diagnosis (and coding) should be made. When the vascular dementia precedes the Alzheimer's disease, it may be impossible to diagnose the latter on clinical grounds. Includes: primary degenerative dementia of the Alzheimer's type Differential diagnosis Consider: a depressive disorder (F30‐F39); delirium (F05.‐); organic amnesic syndrome (F04); other primary dementias, such as in Pick's, Creutzfeldt‐Jakob or Huntington's disease (F02.‐); secondary dementias associated with a variety of physical diseases, toxic states, etc. (F02.8); mild, moderate or severe mental retardation (F70‐F72). Dementia in Alzheimer's disease may coexist with vascular dementia (to be coded F00.2), as when cerebrovascular episodes (multi‐infarct phenomena) are superimposed on a clinical picture and history suggesting Alzheimer's disease. Such episodes may result in sudden exacerbations of the manifestations of dementia. According to postmortem findings, both types may coexist in as many as 10‐15% of all dementia cases. F00.0 Dementia in Alzheimer's disease with early onset Dementia in Alzheimer's disease beginning before the age of 65. There is relatively rapid deterioration, with marked multiple disorders of the higher cortical functions. Aphasia, agraphia, alexia, and apraxia occur relatively early in the course of the dementia in most cases. Diagnostic guidelines As for dementia, described above, with onset before the age of 65 years, and usually with rapid progression of symptoms. Family history of Alzheimer's disease is a contributory but not necessary factor for the diagnosis, as is a family history of Down's syndrome or of lymphoma. Includes: Alzheimer's disease, type 2 presenile dementia, Alzheimer's type F00.1 Dementia in Alzheimer's disease with late onset Dementia in Alzheimer's disease where the clinically observable onset is after the age of 65 years and usually in the late 70s or thereafter, with a slow progression, and usually with memory impairment as the principal feature. Diagnostic guidelines As for dementia, described above, with attention to the presence or absence of features differentiating the disorder from the early‐onset subtype (F00.0). Includes: Alzheimer's disease, type 1 senile dementia, Alzheimer's type F00.2 Dementia in Alzheimer's disease, atypical or mixed type Dementias that do not fit the descriptions and guidelines for either F00.0 or F00.1 should be classified here; mixed Alzheimer's and vascular dementias are also included here.

NINDS – AIREN (Wetterling 1996)	I. The criteria for the clinical diagnosis of probable vascular dementia include all of the following: 1. Dementia defined by cognitive decline from a previously higher level of functioning and manifested by impairment of memory and of two or more cognitive domains (orientation, attention, language, visuospatial functions, executive functions, motor control, and praxis), preferable established by clinical examination and documented by neuropsychological testing; deficits should be severe enough to interfere with activities of daily living not due to physical effects of stroke alone. Exclusion criteria: cases with disturbance of consciousness, delirium, psychosis, severe aphasia, or major sensorimotor impairment precluding neuropsychological testing. Also excluded are systemic disorders or other brain diseases (such as AD) that in and of themselves could account for deficits in memory and cognition. 2. Cerebrovascular disease, defined by the presence of focal signs on neurologic examination, such as hemiparesis, lower facial weakness, Babinski sign, sensory deficit, hemianopia, and dysarthria consistent with stroke (with or without history of stroke), and no evidence of CVD by brain imaging (CT or MRI) including multiple large vessel infarcts or a single strategically placed infarct (angular gyrus, thalamus, basal forebrain, or PCA or ACA territories), as well as multiple basal ganglia and white matter lacunes, or extensive periventricular white matter lesions, or combinations thereof. 3. A relationship between the above two disorders, manifested or inferred by the presence of one or more of the following: (a) onset of dementia within 3 months following a recognized stroke; (b) abrupt deterioration in cognitive functions; or fluctuating, stepwise progression of cognitive deficits.
	II. Clinical features consistent with the diagnosis of probable vascular dementia include the following: (a) Early presence of gait disturbance (small‐step gait or marche a petits pas, or magnetic, apraxic‐ataxic or parkinsonian gait); (b) history of unsteadiness and frequent, unprovoked falls; (c) early urinary frequency, urgency, and other urinary symptoms not explained by urologic disease; (d) pseudobulbar palsy; and (e) personality and mood changes, abulia, depression, emotional incontinence, or other subcortical deficits including psychomotor retardation and abnormal executive function.
	III. Features that make the diagnosis of vascular dementia uncertain or unlikely include: (a) early onset of memory deficit and progressive worsening of memory deficit and progressive worsening of memory and other cognitive functions such as language (transcortical sensory aphasia), motor skills (apraxia), and perception (agnosia), in the absence of corresponding focal lesions on brain imaging; (b) absence of focal neurological signs, other than cognitive disturbance; and (c) absence of cerebrovascular lesions on brain CT or MRI.
	IV. Clinical diagnosis of possible vascular dementia may be made: · in the presence of dementia (section I‐1) with focal neurologic signs in patients in whom brain imaging studies to confirm definite CVD are missing; or · in the absence of clear temporal relationship between dementia and stroke; · or in patients with subtle onset and variable course (plateau or improvement) of cognitive deficits and evidence of relevant CVD.
	V. Criteria for diagnosis of definite vascular dementia are: (a) clinical criteria for probable vascular dementia; (b) histopathologic evidence of CVD obtained from biopsy or autopsy; (c) absence of neurofibrillary tangles and neuritic plaques exceeding those expected for age; and (d) absence of other clinical or pathological disorder capable of producing dementia.
	VI. Classification of vascular dementia for research purposes may be made on the basis of clinical, radiologic, and neuropathologic features, for subcategories or defined conditions such as cortical vascular dementia, subcortical vascular dementia, BD, and thalamic dementia. The term “AD with CVD” should be reserved to classify patients fulfilling the clinical criteria for possible AD and who also present clinical or brain imaging evidence of relevant CVD. Traditionally, these patients have been included with VaD in epidemiologic studies. The term “mixed dementia,” used hitherto, should be avoided.
NINDS – AIREN (Roman 1993)	Diagnosis of Probable VD 1. Dementia Impairment of memory / Impairment of memory and ≥2 cognitive domains / Orientation / Attention / Language / Visuospatial functions / Executive functions, motor control, and praxis / Dementia according to NINDS‐AIREN criteria 2. Cerebrovascular disease Focal signs on neurological examination (hemiparesis, lower facial weakness, Babinski’s sign, sensory deficit, hemianopia, and dysarthria) / Evidence of relevant cerebrovascular disease by brain imaging (CT) / Large‐vessel infarcts / Single strategically placed infarct / Multiple basal ganglia and white matter lacunes / Extensive periventricular white matter lesions / Combinations thereof 3. A relationship between the above disorders manifested or inferred by the presence of ≥1 of the following / Onset of dementia within 3 mo after a recognized stroke / Abrupt deterioration in cognitive functions / Fluctuating, stepwise progression of cognitive deficits 4. Clinical features consistent with the diagnosis of probable VD Early presence of a gait disturbance / History of unsteadiness or frequent, unprovoked falls / Early urinary incontinence / Pseudobulbar palsy / Personality and mood changes 5. Features that make the diagnosis of VD uncertain Early onset of memory deficit and progressive worsening of memory and other cognitive functions in the absence of focal neurological signs and cerebrovascular lesions on CT or MRI
DSM‐IV‐TR (APA 2000)	Diagnosis of vascular dementia A. The development of multiple cognitive deficits manifested by both: 1. Memory impairment (impaired ability to learn new information or to recall previously learned information) 2. One or more of the following cognitive disturbances: (a) aphasia (language disturbance) (b) apraxia (impaired ability to carry out motor activities despite intact motor function) (c) agnosia (failure to recognize or identify objects despite intact sensory function) (d) disturbance in executive functioning (i.e., planning, organizing, sequencing, abstracting) B. The cognitive deficits in criteria A1 and A2 each cause significant impairment in social or occupational functioning and represent a significant decline from a previous level of functioning. C. Focal neurological signs and symptoms (e.g., exaggeration of deep tendon reflexes, extensor plantar response, pseudobulbar palsy, gait abnormalities, weakness of an extremity) or laboratory evidence indicative of cerebrovascular disease (e.g., multiple infarctions involving cortex and underlying white matter) that are judged to be etiologically related to the disturbance. D. The deficits do not occur exclusively during the course of a delirium.
ICD10 Criteria (WHO 1993)	F01 Vascular dementia Vascular (formerly arteriosclerotic) dementia, which includes multi‐infarct dementia, is distinguished from dementia in Alzheimer's disease by its history of onset, clinical features, and subsequent course. Typically, there is a history of transient ischemic attacks with brief impairment of consciousness, fleeting pareses, or visual loss. The dementia may also follow a succession of acute cerebrovascular accidents or, less commonly, a single major stroke. Some impairment of memory and thinking then becomes apparent. Onset, which is usually in later life, can be abrupt, following one particular ischemic episode, or there may be more gradual emergence. The dementia is usually the result of infarction of the brain due to vascular diseases, including hypertensive cerebrovascular disease. The infarcts are usually small but cumulative in their effect. Diagnostic guidelines The diagnosis presupposes the presence of a dementia as described above. Impairment of cognitive function is commonly uneven, so that there may be memory loss, intellectual impairment, and focal neurological signs. Insight and judgement may be relatively well preserved. An abrupt onset or a stepwise deterioration, as well as the presence of focal neurological signs and symptoms, increases the probability of the diagnosis; in some cases, confirmation can be provided only by computerized axial tomography or, ultimately, neuropathological examination. Associated features are: hypertension, carotid bruit, emotional lability with transient depressive mood, weeping or explosive laughter, and transient episodes of clouded consciousness or delirium, often provoked by further infarction. Personality is believed to be relatively well preserved, but personality changes may be evident in a proportion of cases with apathy, disinhibition, or accentuation of previous traits such as egocentricity, paranoid attitudes, or irritability. Includes: arteriosclerotic dementia Differential diagnosis. Consider: delirium (F05.‐); other dementia, particularly in Alzheimer's disease (F00.‐); mood [affective] disorders (F30‐F39); mild or moderate mental retardation (F70‐F71); subdural haemorrhage (traumatic (S06.5), nontraumatic (162.0)). Vascular dementia may coexist with dementia in Alzheimer's disease (to be coded F00.2), as when evidence of a vascular episode is superimposed on a clinical picture and history suggesting Alzheimer's disease. A. Evidence of each of the following 1. Decline in memory (mainly short‐term memory) 2. Decline in other cognitive abilities Deficits in criterion A cause a significant impairment of social functioning B. Absence or clouding of consciousness C. Decline in emotional control or motivation or a change in social behaviour D. Symptoms in criterion A have been present ≥6 months Dementia according to DCR‐10 criteria Unequal distribution of deficits in higher cognitive functions Evidence of focal brain damage Evidence of cerebrovascular disease
ADDTC Criteria (Chui 1992)	Diagnosis of Probable Ischemic VD 1. Dementia (as defined in the text) 2. History, neurological signs, and/or Neuroimaging studies (CT or T₁‐weighted MRI), or Occurrence of a single stroke with a clearly documented temporal relationship to the onset of dementia 3. Evidence of ≥1 infarct outside the cerebellum by CT or T₁‐weighted MRI B. Diagnosis of probable IVD is supported by 1. Evidence of multiple infarcts in brain regions known to affect cognition (as defined by NINDS‐AIREN criteria) 2. History of multiple transient ischemic attacks 3. History of vascular risk factors (e.g., hypertension, heart disease, diabetes mellitus) 4. Elevated Hachinski Ischemia Scale score (≥7) C. Clinical features that are thought to be associated with ischemic VD but await further research 1. Relatively early appearance of gait disturbance and urinary incontinence 2. Periventricular and deep white matter changes on T₂‐weighted MRI that are excessive for age 3. Focal changes in electroencephalographic studies D. Other clinical features that do not constitute strong evidence either for or against a diagnosis of probable ischemic VD 1. Periods of slowly progressive symptoms 2. Illusions, psychoses, hallucinations, delusions 3. Seizures E. Clinical features that cast doubt on a diagnosis of probable ischemic VD 1. Transcortical sensory aphasia in the absence of corresponding focal lesions on neuroimaging studies 2. Absence of central neurological symptoms/signs other than cognitive disturbance

Source	Search strategy	Hits retrieved
1. MEDLINE In‐process and other non‐indexed citations and MEDLINE 1946‐present (Ovid SP) [Date of most recent search: 18 February 2020]	1. exp Dementia/ 2. Cognition Disorders/ 3. exp Neurofibrils/ 4. Neurofilament Proteins/ 5. Senile Plaques/ 6. Neuropil Threads/ 7. (alzheimer$ or dement$).ti,ab. 8. (neurofibril$ adj3 tangle$).ti,ab. 9. (neurofilament adj3 protein$).ti,ab. 10. ((senile or amyloid or neuritic) adj3 plaque$).ti,ab. 11. (neuropil adj3 thread$).ti,ab. 12. or/1‐11 13. exp Amyloid Beta‐Protein/ 14. Peptide Fragments/ 15. ABPP.ti,ab. 16. APP.ti,ab. 17. beta?A4.ti,ab. 18. (beta adj3 A4).ti,ab. 19. Abeta$.ti,ab. 20. amyloid.ti,ab. 21. (amyloidogenic adj3 (peptide$ or protein$)).ti,ab. 22. (Innotest or Inno‐bia or Alzbio3).ti,ab. 23. or/13‐22 24. 12 and 23 25. (cerebrospinal fluid$ or csf or spinal fluid$).ti,ab. 26. (blood or plasma).ti,ab. 27. Cerebrospinal Fluid/ 28. Blood‐Brain Barrier/ 29. or/25‐28 30. (cf or bl or di or du).fs. 31. 29 or 30 32. 24 and 31 33 Cerebrospinal Fluid Proteins/ 34 Biological Markers/cf, bl [Cerebrospinal Fluid, Blood] 35 33 or 34 36 1 and 35 37 32 or 36 38 exp Animals/ not Humans.sh. 39 37 not 38	Mar 2011: 4836 Jan 2012: 707 Jan 2013: 849 Mar 2016: 2414 Jun 2017: 1329 Jan 2018: 806 Nov 2018: 787 Feb 2020: 1211 Total: 12939
2. Embase 1980‐2020 February 17 (Ovid SP) [Date of most recent search: 18 February 2020]	1. exp dementia/ 2. exp cognitive defect/ or exp mild cognitive impairment/ 3. exp neurofilament/ 4. exp neurofilament protein/ 5. senile plaque/ 6. neuropil thread/ 7. (alzheimer$ or dement$).ti,ab. 8. ((cognit$ or memory or cerebr$ or mental$) adj3 (declin$ or impair$ or los$ or deteriorat$ or degenerat$ or complain$ or disturb$ or disorder$)).ti,ab. 9. (neurofibril$ adj3 tangle$).ti,ab. 10. (neurofilament adj3 protein$).ti,ab. 11. ((senile or amyloid or neuritic) adj3 plaque$).ti,ab. 12. (neuropil adj3 thread$).ti,ab. 13. exp amyloid beta protein/ 14. peptide fragment/ 15. ABPP.ti,ab. 16. APP.ti,ab. 17. beta?A4.ti,ab. 18. Abeta$.ti,ab. 19. amyloid.ti,ab. 20. (beta adj3 A4).ti,ab. 21. (amyloidogenic adj3 (peptide$ or protein$)).ti,ab. 22. (Innotest or Inno‐bia or Alzbio3).ti,ab. 23. (cerebrospinal fluid$ or csf or spinal fluid$).ti,ab. 24. (blood or plasma).ti,ab. 25. cerebrospinal fluid/ 26. blood brain barrier/ 27. protein cerebrospinal fluid level/ 28. biological marker/ and (blood or plasma or CSF or "cerebrospinal fluid").ti,ab. 29. or/1‐12 30. or/13‐22 31. or/23‐26 32. 29 and 30 and 31 33. animal/ not (human/ and animal/) 34. 32 not 33 35. review.ti. 36. 34 not 35 37. review.pt. 38. 36 not 37	Mar 2011: 4098 Jan 2012: 1159 Jan 2013: 1062 Mar 2016: 4277 Jun 2017: 3339 Jan 2018: 1423 Nov 2018: 2112 Feb 2020: 3074 Total: 20,544
3. PsycINFO 1806‐February week 3 2020 (Ovid SP) [Date of most recent search: 18 February 2020]	1. exp Dementia/ 2. exp Cognitive Impairment/ 3. Neurofibril*.mp. 4. exp Neurofibrillary Tangles/ 5. Senile Plaques/ 6. "neuropil threads".mp. 7. (alzheimer$ or dement$).ti,ab. 8. ((cognit$ or memory or cerebr$ or mental$) adj3 (declin$ or impair$ or los$ or deteriorat$ or degenerat$ or complain$ or disturb$ or disorder$)).ti,ab. 9. (neurofibril$ adj3 tangle$).ti,ab. 10. (neurofilament adj3 protein$).ti,ab. 11. ((senile or amyloid or neuritic) adj3 plaque$).ti,ab. 12. (neuropil adj3 thread$).ti,ab. 13. exp Beta Amyloid/ 14. exp Peptides/ 15. ABPP.ti,ab. 16. APP.ti,ab. 17. beta?A4.ti,ab. 18. (beta adj3 A4).ti,ab. 19. Abeta$.ti,ab. 20. amyloid.ti,ab. 21. (amyloidogenic adj3 (peptide$ or protein$)).ti,ab. 22. (Innotest or Inno‐bia or Alzbio3).ti,ab. 23. (cerebrospinal fluid$ or csf or spinal fluid$).ti,ab. 24. (blood or plasma).ti,ab. 25. Cerebrospinal Fluid/ 26. Blood Brain Barrier/ 27. Cerebrospinal Fluid/ 28. exp Biological Markers/ 29. or/1‐12 30. or/13‐21 31. or/22‐28 32. 29 and 30 and 31 33. 28 and 29 34. 32 or 33	Mar 2011: 2054 Jan 2012: 468 Jan 2013: 477 Mar 2016: 1732 Jun 2017: 1329 Jan 2018: 670 Nov 2018: 463 Feb 2020: 741 Total: 7934
4. BIOSIS Previews (Thomson Reuters Web of Science) [Date of most recent search: 18 February 2020]	Topic=(dement* OR alzheimer* OR neurofibril* OR neurofilament* OR "senile plaques" OR "neuropil thread") AND Topic=("Amyloid Beta" OR "peptide fragment" OR ABPP OR APP OR abeta OR (beta AND amyloid) OR amyloid OR ab42 OR ab40 OR ("amino acid" AND (40 OR 42)) OR Innotest OR "Inno‐bia" OR Alzbio3) AND Topic=("cerebrospinal fluid" OR csf OR "spinal fluid" OR blood OR plasma) Timespan=All Years. Databases=BIOSIS Previews. Lemmatization=On	Mar 2011: 1311 Jan 2012: 985 Jan 2013: 826 Mar 2016: 1302 Jun 2017: 324 Jan 2018: 291 Nov 2018: 185 Feb 2020: 1104 Total: 6328
5. Web of Science Core Collection, including Conference Proceedings Citation Index (Thomson Reuters Web of Science) (1945‐present) [Date of most recent search: 18 February 2020]	Topic=(dement* OR alzheimer* OR neurofibril* OR neurofilament* OR "senile plaques" OR "neuropil thread") AND Topic=("Amyloid Beta" OR "peptide fragment" OR ABPP OR APP OR abeta OR (beta AND amyloid) OR amyloid OR ab42 OR ab40 OR ("amino acid" AND (40 OR 42)) OR Innotest OR "Inno‐bia" OR Alzbio3) AND Topic=("cerebrospinal fluid" OR csf OR "spinal fluid" OR blood OR plasma) Timespan=All Years. Databases=BIOSIS Previews. Lemmatization=On	Mar 2011: 1764 Jan 2012: 712 Jan 2013: 780 Mar 2016: 1710 Jun 2017: 1117 Jan 2018: 1278 Nov 2018: 1179 Feb 2020: 1458 Total: 9998
6. LILACS (BIREME) [Date of most recent search: 18 February 2020]	“peptídeo beta‐amilóide” OR “placas neuríticas” OR “emaranhados neurofibrilares” OR “senile plaques” OR “β‐amyloid” OR “beta‐amiloide” OR “b‐Amiloid” OR “ovillos neurofibrilares” OR amilóide OR innotest OR “inno‐bia” OR alzbio3 [Words] and CSF OR LCR OR cefalorraquidiano OR “biological marker” OR “biological markers” OR plasma OR plasmáticos OR plasmocitos [Words] and “comprometimento cognitivo leve” OR “cognitive impairment” OR MCI OR Alzheimer OR Alzheimer’s OR AD OR memory OR Memória OR memórias OR demências OR demência OR dementia [Words]	Mar 2011: Jan 2012: 6 Jan 2013: 13 Mar 2016: Jun 2017: 1 Jan 2018: 0 Nov 2018: 0 Feb 2020: 0 Total: 20
TOTAL from all searches before de‐duplication		57,763
TOTAL after de‐duplication		34,027
TOTAL after first assessment by team of seven assessors and CDCIG information specialist		1,835

Domain	Patient selection	Index test	Reference standard	Flow and timing
Description	Describe methods of patient selection: Describe included patients (prior testing, presentation, intended use of index test and setting)	Describe the index test and how it was conducted and interpreted	Describe the reference standard and how it was conducted and interpreted	Describe any patients who did not receive the index test(s) and/or reference standard or who were excluded from the 2 x 2 table (refer to flow diagram): Describe the time interval and any interventions between index test(s) and reference standard
Signalling questions (yes/no/unclear)	Was a consecutive or random sample of patients enrolled?	Were the index test results interpreted without knowledge of the results of the reference standard?	Is the reference standard likely to correctly classify the target condition?	Was there an appropriate interval between index test(s) and reference standard?
	Was a case‐control design avoided?	If a threshold was used, was it pre‐specified?	Were the reference standard results interpreted without knowledge of the results of the index test?	Did all patients receive a reference standard?
	Did the study avoid inappropriate exclusions?			Did all patients receive the same reference standard?
	Did the study avoid inappropriate exclusions?			Were all patients included in the analysis?
Risk of bias: High/low/unclear	Could the selection of patients have introduced bias?	Could the conduct or interpretation of the index test have introduced bias?	Could the reference standard, its conduct, or its interpretation have introduced bias?	Could the patient flow have introduced bias?
Concerns regarding applicability: High/low/unclear	Are there concerns that the included patients do not match the review question?	Are there concerns that the index test, its conduct, or interpretation differ from the review question?	Are there concerns that the target condition as defined by the reference standard does not match the review question?

Question	Response and weighting	Explanation
Patient Selection
Was the sampling method appropriate?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Where sampling is used, the designs least likely to cause bias are consecutive sampling or random sampling. Sampling that is based on volunteers or selecting subjects from a clinic or research resource is prone to bias.
Was a case‐control or similar design avoided?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Designs similar to case control that may introduce bias are those designs where the study team deliberately increase or decrease the proportion of subjects with the target condition, which may not be representative. Some case control methods may already be excluded if they mix subjects from various settings.
Are exclusion criteria described and appropriate?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Study will be automatically graded unclear if exclusions are not detailed (pending contact with study authors). Where exclusions are detailed, the study will be graded as "low risk" if exclusions are felt to be appropriate by the review authors. Certain exclusions common to many studies of dementia are: medical instability; terminal disease; alcohol/substance misuse; concomitant psychiatric diagnosis; other neurodegenerative condition. Exclusions are not felt to be appropriate if 'difficult to diagnose' patients are excluded. Post hoc and inappropriate exclusions will be labelled "high risk" of bias.
Index Test
Was plasma and CSF ABeta42 biomarkers' assessment/interpretation performed without knowledge of clinical dementia diagnosis?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Terms such as "blinded" or "independently and without knowledge of" are sufficient and full details of the blinding procedure are not required. Interpretation of the results of the index test may be influenced by knowledge of the results of reference standard. If the index test is always interpreted prior to the reference standard then the person interpreting the index test cannot be aware of the results of the reference standard and so this item could be rated as 'yes'. For certain index tests the result is objective and knowledge of reference standard should not influence result, for example level of protein in cerebrospinal fluid, in this instance the quality assessment may be "low risk" even if blinding was not achieved.
Were plasma and CSF ABeta42 biomarkers' thresholds pre‐specified?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	For scales and biomarkers there is often a reference point (in units or categories) above which subjects are classified as "test positive"; this may be referred to as threshold; clinical cut‐off or dichotomisation point. A study is classified high risk of bias if the authors define the optimal cut‐off post‐hoc based on their own study data because selecting the threshold to maximise sensitivity and specificity may lead to overoptimistic measures of test performance. Certain papers may use an alternative methodology for analysis that does not use thresholds and these papers should be classified as not applicable.
Reference Standard
Is the assessment used for clinical diagnosis of dementia acceptable?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Commonly used international criteria to assist with clinical diagnosis of dementia include those detailed in DSM‐IV and ICD‐10. Criteria specific to dementia subtypes include but are not limited to NINCDS‐ADRDA criteria for Alzheimer's dementia; McKeith criteria for Lewy Body dementia; Lund criteria for frontotemporal dementia; and the NINDS‐AIREN criteria for vascular dementia. Where the criteria used for assessment is not familiar to the review authors or the Cochrane Dementia and Cognitive Improvement group ('unclear') this item should be classified as "high risk of bias".
Was clinical assessment for dementia performed without knowledge of the plasma and CSF ABeta42 biomarkers?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Terms such as "blinded" or "independently and without knowledge of" are sufficient and full details of the blinding procedure are not required. Interpretation of the results of the reference standard may be influenced by knowledge of the results of index test.
Patient flow
Was there an appropriate interval between plasma and CSF ABeta42 biomarkers and clinical dementia assessment? Were all patients who received plasma and/or biomarkers assessment included in the final analysis?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	For a cross‐sectional study design, there is potential for the patient to change between assessments. The ideal would be same day assessment. We have set an arbitrary maximum interval of one month between tests, although this may be revised depending on the test and the stability of the condition of interest. If drop outs these should be accounted for; a maximum proportion of drop outs to remain low risk of bias has been specified as 20%. Weighting: Low risk
Did all subjects get the same assessment for dementia regardless of plasma and CSF ABeta42 biomarkers?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	There may be scenarios where subjects who score "test positive" on index test have a more detailed assessment. Where dementia assessment differs between subjects this should be classified as high risk of bias.
Were all patients who received plasma and CSF ABeta42 biomarker's assessment included in the final analysis?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	If the number of patients enrolled differs from the number of patients included in the 2 x 2 table then there is the potential for bias. If patients lost to drop‐outs differ systematically from those who remain, then estimates of test performance may differ. If drop outs these should be accounted for; a maximum proportion of drop outs to remain low risk of bias has been specified as 20%
Were missing plasma and CSF ABeta42 biomarkers' results or uninterpretable plasma and CSF ABeta42 biomarkers' biomarker results reported?	No = high risk of bias Yes = low risk of bias Unclear = unclear risk of bias	Where missing or uninterpretable results are reported, and if there is substantial attrition (we have set an arbitrary value of 50% missing data), this should be scored as 'no'. If those results are not reported, this should be scored as 'unclear' and authors will be contacted.
Anchoring statements to assist with assessment for applicability
Question	Explanation
Were included patients representative of the general population of interest?	The included patients should match the intended population as described in the review question. The review authors should consider population in terms of symptoms; pre‐testing; potential disease prevalence; setting. If there is a clear ground for suspecting an unrepresentative spectrum the item should be rated poor applicability.
Index test
Were sufficient data on plasma and CSF ABeta42 biomarkers' application given for the test to be repeated in an independent study?	Variation in technology, test execution, and test interpretation may affect estimate of accuracy. In addition, the background, and training/expertise of the assessor should be reported and taken in consideration. If plasma and CSF ABeta42 biomarkers were not performed consistently this item should be rated poor applicability.
Reference standard
Was clinical diagnosis of dementia made in a manner similar to current clinical practice?	For many reviews, inclusion criteria and assessment for risk of bias will already have assessed the dementia diagnosis. For certain reviews an applicability statement relating to reference standard may not be applicable. There is the possibility that a form of dementia assessment, although valid, may diagnose a far larger proportion of subjects with disease than usual clinical practice. In this instance the item should be rated poor applicability.

Test	No. of studies	No. of participants
1 CSF ABeta42 ADD vs non‐ADD (all studies)	13	1704
2 CSF ABeta42 ADD vs non‐ADD (threshold ≤ 500 pg/ml)	7	1160
3 CSF ABeta42 ADD vs non‐ADD (threshold > 500 pg/ml)	5	406
4 CSF ABeta42 ADD vs VaD (all studies)	11	1151
5 CSF ABeta42 AD vs VaD (threshold ≤ 500 pg/ml)	7	809
6 CSF ABeta42 ADD vs VaD (threshold > 500 pg/ml)	3	194
7 CSF ABeta42 ADD vs FTD (all studies)	17	1948
8 CSF ABeta42 ADD vs FTD (threshold ≤ 500 pg/ml)	8	1033
9 CSF ABeta42 vs FTD (threshold > 500 pg/ml)	5	513
10 CSF ABeta42 ADD vs DLB (all studies)	9	1929
11 CSF ABeta42 ADD vs DLB (threshold ≤ 500 pg/ml)	6	751
12 CSF ABeta42 ADD vs NPH	4	336
13 CSF ABeta42 ADD vs CJD	3	382
14 CSF ABeta42 ADD vs ARCD	1	53
15 CSF ABeta42 ADD vs bvFTD	8	898
16 CSF ABeta42 ADD vs PPA	3	192

*Study characteristics*
Patient Sampling	Retrospective analysis of CSF samples at the Institute of Neurological Sciences of Bologna obtained between 2005 and 2016. Samples were taken from patients with a clinical, genetic, or pathologically confirmed diagnosis of FTD or ADD, and cognitively healthy controls. A sub‐sample of 141 FTD patients were selected who did not have co‐existing DLB, ADD, prion diease, or vascular dementia. Sampling procedure: not reported. Separate data were available for the performance of biomarkers in distinguishing between ADD from FTD. We did not include data on performance of the index test to discriminate ADD participants from controls. Exclusion criteria: patients with CBS were excluded, as were those with significant cerebrovascular pathology on brain imaging. DLB was excluded clinically. No other exclusion criteria were detailed.
Patient characteristics and setting	The sample considered in the review comprised of 201 participants, 60 ADD and 141 FTD. All participants underwent clinical history, neurological examination, neuropsychological testing, and neuroimaging. In addition, some participants had post‐mortem diagnoses and results from molecular genetic testing. Education, gender, and age at the time of lumbar puncture were similar in ADD and FTD. MMSE score was lower in ADD (p < 0.05). Sex: 33 males, 27 females for ADD; 75 males and 66 females for FTD Age mean (SD) (y): 67.1±8.7 for ADD; 64.9 ±9.8 for FTD MMSE: 20.7±4.8 for ADD; 25.0±3.7 for FTD Disease duration (y): not reported Education (y): 10.8±4.8 for ADD; 8.9±4.0 for FTD Sources of recruitment: CSF samples submitted for analysis at the Institute of Neurological Sciences of Bologna
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: >482 ng/L; not prespecified; determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [No]
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis of ADD from FTD) Reference standards: International Working Group 2 (IWG‐2) criteria for ADD and CSF biomarker profile. FTD were classified using criteria for the following subtypes: behavioural variant, non‐fluent variant of primary progressive aphasia, sematic variant of primary progressive aphasia, amyotrophic lateral sclerosis, corticobasal syndrome, progressive supranuclear palsy and FTD with parkinsonism. FTD was neuropathologically confirmed in four cases, and 22 cases had additional molecular genetic findings which supported the diagnosis. Ten participants with FTD were excluded where the CSF biomarker profile was in‐keeping with a diagnosis of ADD. The final clinical diagnosis was confirmed after at least two years of follow‐up. The reference standard results were reported using knowledge of the results of index test.
Flow and timing	The final clinical diagnosis was established after 24 months of follow‐up. AD vs FTD (n=201) AD=60; bvFTD=53; Sensitivity=89%; Specificity=80% (Table 2, p381) TP=53; FP=11; FN=7; TN=42 (calculated in RevMan5) Missing data: Data were requested from the author on the bvFTD subtype and ADD. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	No
Could the reference standard, its conduct, or its interpretation have introduced bias?		High risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	A study with retrospective design (retrospective analysis) of data from patients with DLB and ADD. Consecutive patients with clinical diagnosis of DLB, who were referred to either the movement disorder clinic of the Department of Neurology or the memory clinic of the Department of Geriatric Medicine at the Radboud University Nijmegen Medical Centre, and who underwent a lumbar puncture between December 2003 and June 2008, were included. Out of 93 eligible ADD patients from the memory clinic database, an age and gender matched group of 45 ADD patients was randomly drawn. Exclusion criteria: not reported.
Patient characteristics and setting	The sample considered in the review comprised of 68 participants, 45 ADD and 23 DLB. Disease duration, gender, and age at the time of lumbar puncture were similar in AD and DLB. MMSE score was lower in AD (p < 0.05). Sex: 34 males and 11 females for ADD; 18 males and 5 females for DLB Age mean (SD) (y): 71.6±9.4 for ADD; 71.6 ±9.4 for DLB Disease duration (months): 33.0 for ADD; 38.8 for DLB Sources of recruitment: memory clinic and movement disorder clinic, the Radboud University Nijmegen Medical Centre, The Netherlands
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 4 weeks). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: >482 ng/L; not prespecified; determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of ADD from DLB) Reference standards: NINCDS‐ADRDA criteria for ADD. Clinical diagnosis of DLB was based on McKeith criteria. Initial clinical diagnosis was established by a multidisciplinary team consisting of a geriatrician, a neurologist a neuropsychologist prior CSF sample. The final clinical diagnosis was reassessed by a single rater after a follow‐up period of 12 months or longer. Not reported whether the reference standard results were reported without knowledge of the results of index test.
Flow and timing	The final clinical diagnosis was established (reassessed) 12 months or longer after CSF sampling. AD vs DLB (n=65) AD=44; DLB=21; Sensitivity=62%; Specificity=65% (Table 2, p381) TP=13; FP=15; FN=8; TN=29 (calculated in RevMan5) Missing data: CSF Abeta42 sample was unavailable from 2 DLB and 1 AD participants (Total: 23 DLB and 44 ADD, p379)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Yes
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Unclear
Could the reference standard, its conduct, or its interpretation have introduced bias?		Unclear risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Unclear
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	No
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	Participants were recruited at the Dementia clinic, Neurology Department of Coimbra University Hospital. All patients were followed for two years after which the clinical diagnosis was revised.
Patient characteristics and setting	The sample considered in the review comprised of 214 participants, 107 ADD and 107 FTD. Age of onset, gender, and age at the time of lumbar puncture were similar in AD and FTD. MMSE score was lower in AD (p < 0.005). Sex: 37 males and 70 females for ADD; 47 males and 60 females for FTD Age mean (SD) (y): 64.4 ±9.5 for ADD; 66.3 ±9.0 for FTD Age of onset (years): 62.0 ± 9.6for ADD; 62.6± 9.0for FTD Sources of recruitment: Dementia Clinic, Neurology Department of Coimbra University Hospital
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 538pg/ml, not prespecified; determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of ADD from FTD) Reference standards: NINCDS‐ADRDA criteria and McKhann et al for ADD. Clinical diagnosis of FTD was based on the Lund and Manchester clinical criteria. The reference standard results were reported without knowledge of the results of index test.
Flow and timing
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Yes
Was a case‐control design avoided?	Unclear
Did the study avoid inappropriate exclusions?	Yes
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			High
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Unclear
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	Prospective investigation of participants with probable AD, probable DLB and non‐demented disease controls from initially consecutively referred sample to a laboratory for neurochemical evaluation. Separate data were available for the performance of biomarkers in distinguishing between AD from DLB. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: not reported. Exclusion criteria were only reported for the control group.
Patient characteristics and setting	The sample considered in the review comprised of 43 participants, 18 AD and 25 DLB. CSF was collected from hospitalised DLB patients from a clinic specialising in the diagnosis and treatment of Parkinson's disease. CSF of AD patients came from a memory clinic. The mean age and the mean MMSE score did not significantly differ between AD and DLB participants. Sex: 5 males and 13 females for AD; 21 males and 4 females for DLB Age mean (SD) (y): 69.7 ± 10.6 for AD; 72.0 ± 7.5 for DLB Disease duration (y): not reported Sources of recruitment: AD patients from the memory clinic, University of Goettingen; DLB patients: inpatients from a Paracelsus‐Elena Klinic, Kassel; Germany
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 2 days). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 475pg/ml, not prespecified; determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Yes]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD from DLB) Reference standards: NINCDS‐ADRDA and DSM‐IV criteria for AD. Clinical diagnosis of DLB was based on McKeith and DSM‐IV criteria. Diagnosis was established by a psychiatrist and a neurologist (blinded to biomarker results) thorough anamnesis, clinical examination, results of neuropsychological assessment, clinical records of the patients and the best clinical judgement.
Flow and timing	The interval between established clinical diagnosis and blood sample collection was not reported.However, it appears that CSF samples were collected short after establishing the clinical diagnosis of AD and DLB. At baseline: 18 AD; 25 DLB Sample included in the analysis: 18 AD; 23 DLB AD vs DLB (n=41) Disease⁺: 18; Disease^‐: 23 Sensitivity=50%; Specificity=96% (Calculated in Revman5) TP=9; FP=1; FN=9; TN=22 (calculated in RevMan5) Missing data: CSF Abeta42 sample was unavailable from 2 DLB participants (p1772)
Comparative
Notes	Author contacted: there is some discrepancy between our findings and findings data reported in the Table 2, p1775. No reply.
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	A total of 90 patients (30 ADD; 30 FTLD; 30 non‐demented disease controls) were selected on wards and the dementia outpatient clinic of the Universitiy of Goettingen and the dementia outpatient clinic of the Universitiy of Erlangen between 2000 and 2004. Sampling procedure: not reported. Separate data were available for the performance of biomarkers in distinguishing between ADD from FTLD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: not reported
Patient characteristics and setting	The sample considered in the review comprised of 60 participants, 30 ADD and 30 FTLD. 30 non‐demented disease controls were not included. Diagnosis was established by a psychiatrist and a neurologist (blinded to biomarker results), all highly experienced in clinical differential diagnosis of dementias, on the basis of thorough anamnesis, clinical examination, results of neuropsychological assessment, clinical records of the patients and the best clinical judgement Sex: 13 males and 17 females for ADD; 21 males and 9 females for FTLD Age mean (SD) (y): 65.4 ± 7.3 for ADD; 61.6 ± 11.5 for FTLD. The mean age did not significantly differ between those two groups. MMSE: 19.3 ± 5.4 for ADD; 20.7 ± 8.9 for FTLD (for 26 participants). The mean age did not significantly differ between those two groups. Disease duration (y): not reported Sources of recruitment: mixed setting: the wards and the dementia outpatient clinic of the Univerity of Goettingen; 5 AD patients were recruited from the dementia outpatient clinic of the Universitiy of Erlangen; Germany
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 2 days). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: not reported; determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Yes]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD from FTLD) Reference standards: NINCDS‐ADRDA and DSM‐IV criteria for ADD. Diagnosis for FTLD was established on the McKhann 2001 and Neary 1988 criteria. Clinicians were blinded to biomarker results.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	No
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	A total of 151 patients were selected between January 2013 and January 2015. Sampling procedure: Not reported Separate data were available for the performance of biomarkers in distinguishing probable AD and probable DLB as well as mixed diagnosis of ADD and DLB with the other diagnostic groups. In accordance with inclusion criteria in the current review we only included data to differentiate between ADD and DLB with dementia diagnoses.
Patient characteristics and setting	The sample considered in the review comprised of 51 participants, 31 ADD and 20 DLB. Diagnosis was established double‐blinded to biomarker results by clinicians and the biologist. Sex: 12 males and 19 females for ADD; 14 males and 6 females for DLB Age mean (SD) (y): 67.2±9.3 for ADD; 68.8±9.7 for DLB. MMSE: 20.2±4.7 for ADD; 21±4.7 for DLB . Disease duration (y): not reported Sources of recruitment: The tertiary memory clinic of Strasbourg University Hospital
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 500ng/L, pre‐specified Were the index test results reported without knowledge of the reference standard? Not reported
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD from DLB) Reference standards: McKhann's criteria and Duboi's criteria for ADD. Diagnosis for DLB was established on the McKeith's and DSM‐V criteria. Clinicians were blinded to biomarker results.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Unclear
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	No
Were all patients included in the analysis?	Unclear
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	Retrospective multicentre study from six French memory research centres undertaking clinical and biological diagnoses of dementia. All centres used the same diagnostic procedures. Patients were selected from a database between January 2010 and December 2015. 1221 patients were included in the study: 95 control subjects, 57 prodromal‐DLB, 154 DLB with dementia, 132 prodromal‐ADD, and 783 ADD with dementia. Sampling procedure: not reported. Separate data were available for the performance of biomarkers in distinguishing between ADD from DLB. We did not include data on performance of the index test to discriminate AD participants from controls or prodromal syndromes. Exclusion criteria: patients with mixed diagnoses (e.g. ADD and DLB). No other exclusion criteria were detailed.
Patient characteristics and setting	The sample considered in the review comprised of 937 participants, 783 ADD and 154 DLB. 95 non‐demented disease controls were not included. All participants underwent physical, neurological, and neuropsychological assessments, laboratory tests, and brain imaging. ADD was diagnosed according to Albert's and Dubois criteria. DLB was diagnosed according to McKeith's and DSM‐V criteria. Sex: 333 males and 450 females for ADD; 93 males and 61 females for DLB Age mean (SD) (y): 67.5 ± 9 for ADD; 70.5 ± 10.5 for DLB. Participants with DLB were significantly older than those with ADD. MMSE: 19.0 ± 5.8 for ADD; 19.2 ± 5.5 for DLB. MMSE score did not differ significantly between ADD and DLB. Disease duration (y): not reported Sources of recruitment: six French memory centres undertaking clinical and biological diagnoses of dementia.
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 4 hours). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: pre‐specified threshold <700ng/L, optimal cut‐offs also determined by ROC curve analysis (</= 606ng/L). Were the index test results reported without knowledge of the reference standard? [Unlcear]
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis of ADD from DLB) Reference standards: Albert's and Dubois criteria for ADD. Diagnosis for DLB was established on the McKeith and DSM‐V criteria. CSF criteria were not used in the diagnosis of ADD but does not state if clinicians were blinded to the biomarker results.
Flow and timing	AD vs FTD (n=937) AD=783; DLB=154; Sensitivity=71%; Specificity=53% (Table 2, p381) TP=556; FP=72; FN=227; TN=81 (calculated in RevMan5) Missing data: None. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	No
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	248 patients (109 AD, 41 VD,15 FTD, 25 MCI and 58 controls) were recruited from the Memory Clinic of the Department of Neurology, University Hospital of Ulm over 3 years. Sample procedure not reported. Separate data were available for the performance of biomarkers in distinguishing between AD and other types of dementia. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: not reported.
Patient characteristics and setting	248 participants were included in the study: 109 AD, 41 VD,15 FTD, 25 MCI and 58 controls. Medical history, neurological, neuropsychiatric, neuroradiological and neuropsychological examinations were obtained. Control group: 34 patients presented with tension‐type headache and showed no evidence of a structural, hemorrhagic or inflammatory lesion; 24 patients fulfilled the criteria of a major depressive disorder. CSF samples were collected over 3 years. Separate data were extractable for the accuracy of biomarkers in distinguishing AD dementia from i) FTD & VD and ii) non‐AD dementia. The sample considered in the review comprised of 165 participants (109 AD, 41 VD,15 FTD). Sex: 39 males and 70 females for AD; 24 males and 17 females for VD; 8 males and 7 females for FTD Age: 71 (43‐88) for AD; 75 (47‐88) for VD; 68 (43‐77) for FTD Disease duration (y): 2 (0.5‐10) for AD; 1.75 (0.5‐9) for VD; 2 (0.5‐4) for FTD Sources of referral: secondary care. Not reported Sources of recruitment: Memory Clinic of the Department of Neurology, University Hospital of Ulm, Germany
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 612ng/L, not pre‐specified, cut‐offs were derived from ROC analysis. Were the index test results reported without knowledge of the reference standard? Not reported
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (1. differential diagnosis of AD from VD & FTD; 2. differential diagnosis of AD from non‐AD dementia) Reference standards: NINCDS‐ADRDA criteria Alzheimer's disease dementia Clinical diagnosis of VD was based on NINDS‐AIREN criteria, of FTD on Neary 1998 criteria, of MCI on Pettersen 1999, prior the results of the index test.
Flow and timing	The interval between established clinical diagnosis and blood sample collection was not reported.However, it appears that CSF samples were collected short after establishing the clinical diagnosis of the participants included in the study. Sample included in the analysis: 109 AD; 56 non‐AD (41 VD; 15 FTD) AD vs non‐AD (n=165) Sensitivity=82%; Specificity=46% (Table 3, p294) TP=89; FP=30; FN=20; TN=26 (calculated in RevMan5) All recruited participants with diagnosed dementia were included in the analysis.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	Participants were recruited at the INRCA hospital Neurology Unit, Ancona, Italy. Participants were included where brain atrophy was present as defined by the Pasquier scale (</=2). 95 participants were included: 55 ADD, 21 FTD, and 20 non‐demented controls. Sampling procedure: not reported. Separate data were available for the performance of biomarkers in distinguishing between ADD from FTD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria were: age <60 years, family history of disease, cerebrovascular accidents, anamnesis of delirium, cognitive decline induced by head injury, recently diagnosed or untreated thyroid disease, vitamin B12 or folic acid deficiency, intoxication with drugs or medications, severe depression (pseudodementia), chromosome 21 trisomy (Down syndrome), neurosyphilis, and human immunodeficiency virus dementia.
Patient characteristics and setting	Participants underwent clinical history, neuropsychological and functional assessments, neuroimaging, and laboratory tests. Sex: 23 males and 32 females for ADD; 9 males and 12 females for FTD. Age mean (SD) (y): 77.3 ± 7.1 for ADD; 72.0 ± 5.8 for FTD. Participants with ADD were significantly older than those with FTD. MMSE: 14.5 ± 6.1 for ADD; 19.0 ± 6.2 for FTD. MMSE score was significantly lower in ADD compared to FTD. Disease duration (y): not reported Sources of recruitment: Participants were recruited at the INRCA hospital Neurology Unit, Ancona, Italy.
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 3 hours). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Fujirebio Inc., Tokyo, Japan. Threshold: not pre‐specified, optimal cut‐offs were calculated. Were the index test results reported without knowledge of the reference standard? [Yes]
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis of ADD from FTD) Reference standards: NIA/AA and IWG‐2 criteria for ADD. FTD was diagnosed according to the EFNS‐ENS Guidelines. Participants with FTD were subclassifed according to criteria for behavioural variant and primary progressive aphasia subtypes. Diagnosis was confirmed after at least 24 months of follow‐up. It was not clear if clinicians were blinded to the results of the index test.
Flow and timing	Data were provided by the author upon request. AD vs FTD (n=76) AD=55; FTD=21; Sensitivity=100%; Specificity=0% (Table 2, p381) TP=55; FP=21; FN=0; TN=0 (calculated in RevMan5) Missing data: None. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	Yes
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			High
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Unclear
Could the reference standard, its conduct, or its interpretation have introduced bias?		Unclear risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	Unclear
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	Patients with mild to moderate AD (n=61) or VD (n=25) were selected from a large database containing 260 patients with cognitive impairment or dementia of various origins (e.g., degenerative, vascular, hereditary, inflammatory, metabolic) who visited an outpatient clinic between 1992 and 2004. Thirty controls, aged >50 years, with no neurological disorder, were also included. We only considered data on performance of the index test to discriminate between patients with AD and VD. Excluded criteria: not reported
Patient characteristics and setting	The sample considered in the review comprised of 86 participants, 61 AD and 25 VD. Separate data were reported for the performance of biomarkers to distinguish between AD and VD. The control group was not included. The mean age did not significantly differ between AD and VD participants. Sex: 25 males, 36 females for AD; 14 males, 11 females for VD Age (SD) (y): 68 (8.8) for AD; 72 (8.4) for VD Sources of recruitment: database of patients from an outpatient clinic, the Radboud University Nijmegen Medical Centre, The Netherlands
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 520pg/mL, not pre‐specified, determined by ROC analysis. Cutoff values with the most optimal combination of sensitivity and specificity to discriminate between these AD and VD groups were calculated. Were the index test results reported without knowledge of the reference standard? Not reported
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD from VD) Reference standards: NINCDS‐ADRDA criteria for AD. Clinical diagnosis of VD was based on NINDS‐AIREN criteria (Roman 1993). Clinical diagnosis was established prior to study entry.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported.However, it appears that CSF samples were collected shortly after establishing the clinical differential diagnosis of AD and VD. Lumbar punctures were performed after written informed consent was obtained from the patient and the patient's legal representatives. Sample included in the analysis: 61 AD; 25 VD AD vs VD (n=86) Sensitivity=82%; Specificity=76% (Table 2, p756) TP=50; FP=6; FN=11; TN=19 (calculated in RevMan5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	The enrolment of patients in this prospective study started in January 2006 and ended in December 2009. All consecutive patients admitted to two tertiary memory clinics with an ambiguous diagnosis of AD or fvFTD according to current research criteria (Neary 1998; McKhann 1984) underwent lumbar puncture as a diagnostic tool. 75 ADD patients and 42 fvFTD patients were enrolled. Exclusion criteria: not reported.
Patient characteristics and setting	The sample considered in the review comprised of 114 participants, 72 ADD and 42 fvFTD. MMSE adjusted score was significantly higher (p = 0.04) in fvFTD than in ADD. Sex: 32 males and 40 females for ADD; 26 males and 16 females for fvFTD Age mean (SD) (y): 67±6.8 for ADD; 69±7.1 for fvFTD Disease duration (months): 24.1±12.6 for ADD; 26.9±15.0 for fvFTD MMSE: 18.3±4.2 for ADD; 25.5±4.8 for fvFTD Sources of referral: not reported Sources of recruitment: two tertiary memory clinics, Department of Neurology, IRCCS MUltimedica and Vita‐Salute S. Raffaele University, Milan, Italy
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed (within 15 days). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 104pg/mL, not pre‐specified, determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? Yes
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of ADD from bvFTD) Reference standards: NINCDS‐ADRDA criteria for ADD. Clinical diagnosis of FTD was based on Neary 1998 criteria. Initial clinical diagnosis, independent of CSF metabolite levels, were established, which during the study were known only to researchers not further involved in the follow‐up. Afterwards, patients were evaluated at 6‐months intervals by three expert neurologists blind to the CSF results, who had to confirm or discard the initial clinical diagnosis. After at least 2 years of follow up, the last clinical diagnosis was considered as the gold standard to be compared with CSF biomarkers.
Flow and timing	The final clinical diagnosis was established (reassessed) at least 2 years of follow up after CSF sampling. ADD vs bvFTD (n=114) ADD=72; bvFTD=42 Sensitivity=82%; Specificity=21% (calculated in RevMan5) TP=59; FP=33; FN13; TN=9 (calculated in RevMan5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Yes
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	No
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	A cross‐sectional study of participants under the age of 65 undergoing assessment at the Alzheimer's Disease and Other Cognitive Disorders Unit at the Hospital Clinic de Barcelona. 138 participants were recruited between 2009 and 2016 with the following diagnoses: 64 ADD, 26 FTD, and 48 healthy controls. Sampling procedure: not reported. Separate data were available for the performance of biomarkers in distinguishing between ADD from FTD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: not detailed.
Patient characteristics and setting	Participants underwent neurological and neuropsychological assessments, and neuroimaging. Sex: 28 males and 36 females for ADD; 14 males and 12 females for FTD. Age mean (SD) (y): 56.6 (54.5‐60.5) for ADD; 60.6 (55.9‐64.7) for FTD. Participants with FTD were significantly older than those with ADD. MMSE: 23 (19‐26.5) for ADD; 26.0 (24.0‐27.0) for FTD. MMSE score was not significantly different in ADD compared to FTD. Disease duration (y): 2.9 (1.61‐3.79) for ADD; 2.88 (1.9‐3.78) for FTD. Sources of recruitment: Participants were recruited at the Alzheimer's Disease and Other Cognitive Disorders Unit at the Hospital Clinic de Barcelona.
Index tests	Patients gave CSF samples. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: pre‐specified at <550 pg/ml and 750 pg/ml, but optimal thresholds were used for analysis. Were the index test results reported without knowledge of the reference standard? [Unclear].
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis of ADD from FTD). Reference standards: NIA/AA for ADD: NIA/AA criteria. All participants with ADD had a typical CSF biomarker pattern. FTD was diagnosed by criteria in two subtpyes: behavioural variant and semantic variant of primary progressive aphasia. It was not clear if clinicians were blinded to the results of the index test.
Flow and timing	Data were provided by the author upon request. AD vs FTD (n=23) AD=18; FTD=5; Sensitivity=100%; Specificity=94% (Table 2, p381) TP=5; FP=1; FN=0; TN=17 (calculated in RevMan5) Missing data: None. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	No
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			High
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Unclear
Could the reference standard, its conduct, or its interpretation have introduced bias?		Unclear risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		High risk

PERMALINK

Plasma and cerebrospinal fluid ABeta42 for the differential diagnosis of Alzheimer's disease dementia in participants diagnosed with any dementia subtype in a specialist care setting

Michelle Kokkinou

Lucy C Beishon

Nadja Smailagic

Anna H Noel-Storr

Chris Hyde

Obioha Ukoumunne

Rosemary E Worrall

Anja Hayen

Meera Desai

Abhishekh Hulegar Ashok

Eleanor J Paul

Aikaterini Georgopoulou

Tiziana Casoli

Terry J Quinn

Craig W Ritchie

Abstract

Background

Objectives

Search methods

Selection criteria

Data collection and analysis

Main results

Authors' conclusions

Plain language summary

Summary of findings

Background

Target condition being diagnosed

Index test(s)

Clinical pathway

Alternative test(s)

Rationale

Objectives

Secondary objectives

Methods

Criteria for considering studies for this review

Types of studies

Participants

Index tests

Target conditions

Reference standards

Search methods for identification of studies

Electronic searches

Searching other resources

Data collection and analysis

Selection of studies

Data extraction and management

Assessment of methodological quality

Statistical analysis and data synthesis

Investigations of heterogeneity

Sensitivity analyses

Assessment of reporting bias

Results

Results of the search

1.

Summary of included studies

Index test

1. Included studies and the index test accuracy at study level.

Target disorder

Spectrum of participants

Clinical setting

Methodological quality of included studies

2.

3.

Findings

Summary of findings 1. Summary of findings table.

CSF ABeta42 for differentiating ADD from non‐ADD

4.

5.

6.

7.

8.

9.

CSF ABeta42 for differentiating ADD from VaD

10.

11.

12.

13.

14.

*Study characteristics*
Patient Sampling	Patients who were referred to either the movement disorders clinic of the department of Neurology or the memory clinic of the Department of Geriatric Medicine at the Radboud University Medical Centre during the period May 1996 to December 2009. Patients who had received a lumbar puncture and had relevant CSF parameters and not included in a previous study were included.
Patient characteristics and setting	The sample considered in the review comprised of, 64 ADD, 14 DLB, 15 VaD and 26 FTD subjects. MMSE findings and disease duration were not available for all patients. Sex: 13 males and 51 females for ADD; 10 males and 4 females for DLB, 10 males and 5 females for VaD and 17 males and 9 females for FTD. Age mean (SD) (y): 73.1 ± 8.3 for ADD; 72.4 ± 8.0 for DLB, 76.5 ± 4.8 for VaD and 61.6 ± 8.4 for FTD. Disease duration (months): 15 ± 15.6 or ADD (n=61); 24 ± 24.0 for DLB (n=6), 17 ± 15 for VaD (n= 12) and 7.3 ± 14 for FTD (n= 12). MMSE: 20 ± 4 for ADD (n= 61); 22 ± 5 for DLB (n=4), 18 ± 3.7 for VaD (n=12) and 18 ± 7.3 for FTD (n= 10).
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 500pg/mL, not pre‐specified, determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? Yes
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD dementia from DLB, VaD and FTD) Reference standard: NINCDS‐ADRDA criteria for AD The clinical diagnosis of DLB was based on McKeith criteria, for VaD on NINDS‐AIREN criteria and for FTD on the Lund and Manchester Groups criteria. It is not stated whether the reference standard was performed before applying the index test.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported. However, it appears that CSF samples were collected short after the diagnosis of dementia was confirmed. Sample included in the analysis: 64 AD; 14 DLB; 26 FTD; 15 VaD
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		Low risk
Are there concerns that the included patients and setting do not match the review question?			Unclear
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Unclear
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	A total of 99 subjects were included in the study: 38 patients with AD, 14 patients with CJD and 47 controls. Sampling procedure not reported. We only considered data on performance of the index test to discriminate between patients with AD and CJD. Exclusion criteria not reported.
Patient characteristics and setting	The sample considered in the review comprised of 52 participants: 38 patients with ADD, 14 patients with CJD. Sex: 15 M, 23 F AD; 7 M, 7F CJD Age (y): 68±10 years AD; 59±4 CJD Disease duration (y): 3.6±2.4 AD; 0.4±0.2 CJD Sources of recruitment: Department of Neurology, Athens National University, Greece. Not reported whether inpatients or outpatients
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 445pg/ml; not prespecified; Receiver operating characteristics (ROCs) curve analysis was used to define the cut off concentrations of tau protein and Aβ₄₂ with the corresponding optimal sensitivity and specificity (Fig 1B, p402). Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD dementia from CJD) Reference standard: NINCDS‐ADRDA criteria for AD The clinical diagnosis of CJD was based on progressive dementia of less than 2 years, periodic sharp wave complexes in the EEG recording, and two of the following: (1) myoclonus, (2) visual or cerebeller symptoms, (3) pyramidal or extrapyramidal tract signs, and ( 4) akinetic mutism. All patients had a positive test for 14‐3‐3 protein, a sensitive marker of the disease. The reference standard was performed before applying the index test. Method of confirming diagnosis was not specified for two patients.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported. However, it appears that CSF samples were collected short after the diagnosis of dementia was confirmed. Sample included in the analysis: 38 AD; 12 CJD AD vs CJD (50) TP=29; FP=7; FN=9; TN=5 (Fig 1B, p402) Sensitivity=76%; Specificity=42% (Calculated in RevMan5) Missing data: from 2 CJD participants. it was not reported whether those two patients with clinical diagnosis of CJD, which were not confirmed either postmortem or by biopsy, were excluded from the analysis.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	Participants from an outpatient clinic diagnosed with AD and non‐AD dementia were followed‐up for at least three years in an effort to ensure the correct diagnosis, and doubtful cases were rejected. 70 patients with dementia (49 AD; 15 non‐AD; 6 VD) were recruited. 49 controls were also included. Sample procedure not reported. Separate data were available for the performance of the biomarkers in distinguishing AD from non‐AD dementia, and AD from VD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: patients with dementia due to metabolic causes and patients with a history of alcohol abuse, MRI infarctions (except VD patients), or B12 deficiency were excluded.
Patient characteristics and setting	The sample considered in the review comprised of 70 participants: 49 AD, 15 non‐AD (6 DLB; 4 FTD; 1 with Parkinson's disease; 2 with progressive supranuclear pulsy; 2 with corticobasal‐ganglionic degeneration) and 6 with VD. All participants had detailed evaluation (medical history, physical and neurological examination, blood tests to exclude metabolic causes of dementia) and MRI. Sex: 31 males and 18 females for AD; 11 males and 4 females for non‐AD dementia; 4 males and 2 females for VD Age (SD) (y): 67.6 ± 9.3 for AD; 61.3 ± 5.1 for non‐AD dementia; 69 ± 4 for VD Sources of recruitment: an outpatient clinic, Athens National University, Greece.
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐70°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 435 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (1. differential diagnosis of AD from non‐AD dementia; 2. differential diagnosis of AD from VD) Reference standards: NINCDS‐ADRDA for AD. Clinical diagnosis of VD was based on NINDS‐AIREN criteria, of DLB and Parkinson's dementia on McKeith criteria, of FTD on Neary 1999 criteria, of progressive supranuclear palsy according on NINDS‐SPSP criteria. Criteria of corticobasal‐ganglionic degeneration were not not specified. Clinical diagnosis was established prior the results of the index test.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported. However, it appears that CSF samples were collected shortly after the clinical diagnosis was established. Sample included in the analysis: 49 AD; 6 VD; 15 non‐AD (6 DLB; 4 FTD; 1 with PD dementia; 2 with progressive supranuclear pulsy; 2 with corticobasal‐ganglionic degeneration) 1. AD vs non‐AD dementia (n=64) Sensitivity=71%; Specificity=80% (Abstract) TP=35; FP=3; FN=14; TN=12 (calculated in Revman5) 2. AD vs VD (n=55) Sensitivity=82%; Specificity=67% (Abstract) TP=40; FP=2; FN=9; TN=4 (calculated in Revman5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Yes
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	A total of 103 subjects were included in the study: 33 patients with AD, 20 patients with ARCD and 50 controls (healthy elderly). ARCD patients were recruited during a two‐year period from a larger pool of 82 detoxified alcoholic subjects. No further details about sampling procedure. Separate data were available for the performance of biomarkers in distinguishing between AD from ACRD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria not reported.
Patient characteristics and setting	The sample considered in the review comprised of 53 participants: were included in the review: 33 with AD and 20 with ARCD, which completed a detoxification program. AD patients were subjected to a detailed evaluation (medical history, physical and neurological examination, computed tomography and/or magnetic resonance imaging and blood tests to exclude metabolic causes of dementia). There was no history of alcohol use or abuse and all had a sufficient follow‐up (for at least two years) to ensure diagnosis. No one of the patients was under any medication for dementia at the time of lumbar puncture. Evaluation of alcohol abuse was made by the Pattern of Abuse tool (Hughes 1980), the section on alcoholism of Composite International Diagnostic Interview (WHO 1990) and the Diagnostic Interview Schedule (Wells 1994). The mean duration of alcohol consumption was 29 years (range 6–40 years). Only 23 of the 83 subjects met the DSM‐IV criteria of alcohol‐induced persisting dementia. Three out of the 23 patients were under the age of 40 years (out of the range of AD patients), and were not included in the study. Sex: 14 M, 19 F AD; 18 M, 2 F ACRD Age: 63±11 years AD; 60±12 ACRD MMSE: 23 (15‐27) AD; 25 (15‐28) ACRD Resources of recruitment: i) in‐patients: Drug and Alcohol Addiction Clinic, Department of Psychiatry, Athens National University, Greece; ii) not reported for AD participants
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐70°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 562 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD dementia from ARCD) Reference standard: NINCDS‐ADRDA criteria Clinical diagnostic criteria for ARCD: the Pattern of Abuse tool (Hughes 980), the section on alcoholism of Composite International Diagnostic Interview (WHO 1990) and the Diagnostic Interview Schedule (Wells 1994). The reference standard was performed before applying the index test.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported.However, it appears that CSF samples were collected short after neuropsychological examination that was performed two months after detoxification for alcohol‐induced dementia. Sample included in the analysis: 33 AD; 20 ARCD AD vs ACRD (53) TP=28; FP=4; FN=5; TN=16 (Fig 1B, p402) Sensitivity=85%; Specificity=80% (Abstract)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	A total of 85 patients and 72 elderly controls were recruited. Sample procedure not described. Separate data were available on biomarkers for differentiating AD and idiopathic normal presure hydrocephalus (iNPH) patients. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: patients with secondary NPH (e.g. following meningitis, hemorrhage, brain tumor or trauma) were excluded
Patient characteristics and setting	The sample considered in the review comprised of 85 participants: 67 with AD and 18 with iNPH. All the patients underwent extensive neuropsychological evaluation in an effort to further reduce the possibility of AD comorbidity. At least a 2‐year follow‐up was required to ensure correct diagnosis. No AD patients were under cholinesterase inhibitor therapy at the time of lumbar puncture Sex: 26 males and 41 females for AD; 11 males and 7 females for AD for iNPH Age (SD) (y): 66 ± 10 for AD; 69 ± 14 for iNPH MMSE:18 (14–22) for AD; 21 (16–26) for iNPH Disease duration (y): 3.2 ± 2.3 for AD; 0.7 ± 0.4 for iNPH Sources of recruitment: specialist care setting, Athens National University, Greece. Not reported whether inpatients or outpatients
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 268 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD dementia from iNPH) Reference standard: NINCDS‐ADRDA criteria for AD Clinical diagnostic criteria for iNPH: the standard classic triad of gait impairment, urinary incontinence and impaired mental function, supported by ventricular dilation in neuroimaging without significant cerebral atrophy, with Evan’s index >0.3 on CT or MRI scan. The reference standard was performed before applying the index test.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported. However, it appears that CSF samples were collected short after establishing the clinical diagnosis of AD and iNPH. Sample included in the analysis: 67 AD; 18 iNPH AD vs iNPH (n=85) Sensitivity=91%; Specificity=44% (Table 2, p171) TP=61 FP=10; FN=6; TN=8 (calculated in RevMan5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	A total of 203 participants (76 AD; 34 FTLD; 93 healthy controls) were prospectively enrolled in the study. No further information on sampling procedure. Separate data were available for the performance of biomarkers in distinguishing between AD from FTD and AD from FTLD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: secondary causes of dementia.
Patient characteristics and setting	110 participants were considered in the review: 76 AD and 34 FTLD (24 FTD; 5 PPA; 5 FTD with motor neuron signs). All patients underwent detailed clinical, neuropsychologic, biochemical, and neuroimaging examination (magnetic resonance imaging in all patients and, additionally, single photon emission computed tomography in all FTLD patients), to exclude secondary causes of dementia and establish the diagnosis. In addition, at least 2‐years follow‐up was available to ensure the correct diagnosis. None of the patients were under cholinesterase inhibitors at the time of lumbar puncture. Sex: 28 males and 48 females for AD; 20 males and 14 females for FTLD Age mean (SD) (y): 66.0 ± 10.0 for AD; 3.1 ± 2.7 for FTLD Disease duration (y): 3.4 ± 2.8 for AD; 61.0 ± 9.0 for FTLD Sources of referral: not reported Sources of recruitment: specialist care setting, Athens National University, Greece. Not reported whether inpatients or outpatients
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 451 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD dementia from FTD, and AD from FTLD) Reference standard: NINCDS‐ADRDA criteria for AD The clinical diagnosis of FTLD was established on Neary 1998 criteria. At least 2‐years follow‐up was available to ensure the correct diagnosis, prior the results of the index test. Disease duration was defined as the time between the onset of the symptom(s) and CSF sampling.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported. However, it appears that CSF samples were collected shortly after establishing the clinical diagnoses. Sample included in the analysis: 76 AD and 34 FTD (FTLD: 24 FTD; 5 PPA; 5 FTD with motor neuron signs) AD vs FTD (FTLD) (N=107) TP=57; FP=9; FN=19; TN=22 (Fig 1b, p49) Sensitivity=75%; Specificity=71% (Calculated in RevMan) Missing data: 3 FTLD were not included in the analysis
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	Analysis of CSF samples from 76 ADD, 74 MCI, 11 FTD, and 45 non‐dementia controls. Participants with MCI were followed‐up for 4‐8 years and 21 converted to AD, 53 remained stable. Recruitment procedure: not specified. Sampling procedure: not specified. Separate data were available for the performance of biomarkers in distinguishing between ADD from FTD. We did not include data on performance of the index test to discriminate AD participants from controls. Exclusion criteria: not detailed.
Patient characteristics and setting	Participants underwent clinical history, cognitive assessment, and neuroimaging. Sex: 29 males and 47 females for ADD; 7 males and 4 females for FTD. Age median (range) (y): 72 (54‐88) for ADD; 66 (50‐75) for FTD. Participants with ADD were significantly older than those with FTD. MMSE: 23.6 ± 4.3 for ADD; 25.20 ± 4.2 for FTD. MMSE score was significantly lower in ADD compared to FTD. Disease duration (y): not specified. Sources of recruitment: not specified.
Index tests	Patients gave CSF samples. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: pre‐specified at <530 ng/L. Were the index test results reported without knowledge of the reference standard? [Unclear].
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis ADD from FTD) Reference standards: NINCDS‐ADRDA and DSM‐IV criteria for ADD. FTD diagnostic criteria not stated. It was not clear if clinicians were blinded to the results of the index test.
Flow and timing	Data were provided by the author upon request. AD vs FTD (n=87) AD=76; FTD=11; Sensitivity=88%; Specificity=91% (Table 2, p381) TP=67; FP=1; FN=9; TN=10 (calculated in RevMan5) Missing data: None. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Unclear
Could the reference standard, its conduct, or its interpretation have introduced bias?		Unclear risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	No
Could the patient flow have introduced bias?		High risk

*Study characteristics*
Patient Sampling	A cross‐sectional study at the memory clinic at Oslo University Hospital, Ullevaal, Norway. 205 patients were referred for diagnostic work‐up between January 2009 and July 2014. 138 participants had a diagnosis of ADD, and 17 were "other dementia". Separate data were available for the performance of biomarkers in distinguishing between ADD from FTD. We did not include data on performance of the index test to discriminate AD participants from MCI or subjective cognitive impairment. Sampling procedure: not reported. Inclusion criteria: CSF biomarkers available. Exclusion criteria: none.
Patient characteristics and setting	Participants underwent clinical history, neuropsychological examination, laboratory tests, neuroimaging. Consensus diagnosis was made by two experienced physicians. Sex: 46.3% of the total sample were female. Age mean (SD): 84.8 ±8.8 for the total sample. MMSE: 23.5 ± 4.1 for ADD; 24.3 ± 3.6 for other dementia. MMSE score was significantly lower in ADD compared to FTD. Disease duration (y): not specified. Sources of recruitment: outpatient memory clinic at the Oslo University Hospital, Ullevall, Norway.
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐20°C and analysed (within 1 day). Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: pre‐specified at >550 ng/L and >700 ng/L. Were the index test results reported without knowledge of the reference standard? [Unclear]
Target condition and reference standard(s)	Target condition: Alzheimer's disease (differential diagnosis ADD from other dementia) Reference standards: no diagnostic criteria specified; by consensus between two experiences physicians. Physicians were blinded to the results of the index test.
Flow and timing	Data were provided by the author upon request. AD vs FTD (n=71) AD=59; FTD=12; Sensitivity=43%; Specificity=35% (Table 2, p381) TP=25; FP=8; FN=34; TN=4 (calculated in RevMan5) Missing data: Yes. The interval between established clinical diagnosis and CSF sample collection was not reported.
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	Yes
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		Unclear risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Unclear risk

*Study characteristics*
Patient Sampling	In total 68 participants were recruited (22 AD; 11 non‐AD; 35 controls). Sampling procedure not reported. Separate data were available on the performance of biomarkers to distinguish between ADD and other types of dementia. We did not include data on performance of the index test to discriminate AD participants from controls. No details of recruitment, or exclusion criteria were reported.
Patient characteristics and setting	The sample considered in the review comprised of 33 participants: 22 AD and 11 non‐AD dementia (5 VD; 1 mixed dementia; 1 subcortical arterial sclerotic encephalopathy; 1 senile dementia of vascular origin; 1 FTD accompanied by Still‐Richardson‐Olszewski syndrome; 1 dementia due to alcohol abuse; 1 dementia of unclear etiology). All subjects underwent clinical examination, routine blood, urine and CSF tests, magnetic resonance imaging or computed tomography and neuropsychological tests when applicable. Sex: 6 males and 16 females for AD; 6 males and 5 females for non‐AD dementia Age (SD) (y): 68 (62–77) for AD; 75 (65–80) for non‐AD dementia MMSE: 14 (12–19) for AD; 22 (21–25) for non‐AD dementia Sources of recruitment: specialist care setting, University of Goetting, Germany. Not reported whether inpatients or outpatients.
Index tests	Patients gave CSF samples. The samples were stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 550 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Yes]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (differential diagnosis of AD from non‐AD dementia) Reference standards: NINCDS‐ADRDA for AD. Clinical diagnosis was established prior the results of the index test.
Flow and timing	The interval between established clinical diagnosis and CSF sample collection was not reported.However, it appears that CSF samples were collected short after establishing the clinical diagnosis. Sample included in the analysis: 21 AD; 11 non‐AD (5 VD; 1 mixed dementia; 1 subcortical arterial sclerotic encephalopathy; 1 SD; 1 FTD; 1 dementia due to alcohol abuse; 1 unspecified) AD vs non‐AD (n=33) Sensitivity=86%; Specificity=82% (Table 2, p275) TP=19; FP=2; FN=3; TN=9 (calculated in RevMan5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	Unclear
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Low concern
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Yes
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Low risk

*Study characteristics*
Patient Sampling	CSF samples archived for research purposes from patients with probable AD, VD, iNHP dementia, Parkinson disease without dementia and controls were selected. Separate data on the performance of biomarkers to distinguish between AD from VD and iNPH dementia have been reported. Sample procedure not reported. Exclusion criteria not reported.
Patient characteristics and setting	CSF samples from 36 participants: 12 ADD, 12 VD and 12 iNPH. Sex: 5 males and 7 females for AD; 4 males and 8 females for VD; 9 males and 3 females for iNPH Age (SD) (y): 71.8 ±1.7 AD; 76.4 ±1.9 for VD; 75.0 ±1.9 for iNPH Sources of recruitment: not reported. Not reported whether the study was conducted in Germany or Austria.
Index tests	Patients gave CSF samples. The samples were collected in polypropylene tubes, centrifuged, aliquoted, and stored at ‐80°C and analysed. Abeta42 was measured using enzyme‐linked immunosorbent assays, obtained from Innogenetics NV, Gent, Belgium. Threshold: 562 pg/ml; not prespecified; Cut‐offs were determined by ROC analysis. Were the index test results reported without knowledge of the reference standard? [Not reported]
Target condition and reference standard(s)	Target condition: Alzheimer's disease dementia (1. differential diagnosis of AD from VD; 2. differential diagnosis of AD from iNPH) Reference standards: NINCDS‐ADRDA criteria for ADD. Clinical diagnosis of VD was based on NINDS‐AIREN and ICD‐10 criteria. Clinical diagnosis of iNPH was based on clinical symptoms (Keifer index), the results of neuroimaging and improvement after CSF withdrawal. Clinical diagnosis was established prior the results of the index test.
Flow and timing	Retrospective analysis. The interval between established clinical diagnosis and CSF sample collection was not reported. Sample included in the analysis: 12 AD, 12 VD; 12 iNPH. AD vs VD (n=24) TP=8; FP=6; FN=4; TN=6 (Fig 1, p277) Sensitivity=67%; Specificity=50% (Calculated in RevMan5) AD vs iNPH (n=24) TP=8; FP=8; FN=4; TN=4 (Fig 1, p277) Sensitivity=67%; Specificity=33% (Calculated in RevMan5)
Comparative
Notes
*Methodological quality*
Item	Authors' judgement	Risk of bias	Applicability concerns
DOMAIN 1: Patient Selection
Was a consecutive or random sample of patients enrolled?	No
Was a case‐control design avoided?	No
Did the study avoid inappropriate exclusions?	Unclear
Could the selection of patients have introduced bias?		High risk
Are there concerns that the included patients and setting do not match the review question?			Unclear
DOMAIN 3: Reference Standard
Is the reference standards likely to correctly classify the target condition?	Yes
Were the reference standard results interpreted without knowledge of the results of the index tests?	Yes
Could the reference standard, its conduct, or its interpretation have introduced bias?		Low risk
Are there concerns that the target condition as defined by the reference standard does not match the question?			Low concern
DOMAIN 4: Flow and Timing
Was there an appropriate interval between index test and reference standard?	Unclear
Were all patients included in the analysis?	Yes
Could the patient flow have introduced bias?		Unclear risk