Dementia is a syndrome due to a brain disease — usually of a chronic or progressive nature — in which there is disturbance of multiple higher cortical functions, including memory, thinking, orientation, comprehension, calculation, learning capacity, language, and judgement. However, consciousness remains unaffected. The impairments of cognitive function are commonly accompanied, and occasionally preceded, by a deterioration in emotional control, social behaviour, motivation, and the impairment is sufficient to interfere with everyday activities. Dementia is a collective of different subtypes distinguished by the underlying pathology. Alzheimer's disease dementia (ADD) is the most common form of dementia and other important pathologies associated with dementia are vascular disease, Lewy bodies, and frontotemporal pathology (WHO 2012).

Dementia is a serious worldwide public health problem, with a prevalence of 4.7% in adults older than 60 years (6.2% and 6.5% in Europe and the Americas, respectively). Due to its prevalence in older people, it is expected that the number of people with dementia will increase dramatically. Consequently, in the year 2050, an expected 115 million people will have dementia. This will result in a considerable economic burden, which currently stands at 1% of the world's Gross National Product (GNP) in direct and indirect costs (WHO 2012). These financial costs are in addition to the devastating personal and social consequences of the condition.

The definition of MCI applies to people without evidence of significant deterioration in activities of daily living, but with subjective memory complaints and cognitive impairment detected by standardised tests. MCI often precedes clinical dementia, but there is no consensus regarding how to operationalise the MCI diagnosis. There are several clinical criteria to define which people have MCI, including the Petersen criteria or Petersen Revisited Criteria (Petersen 1999; Petersen 2004; Winblad 2004), Clinical Dementia Rating Scale (CDR= 0.5) (Morris 1993), or 16 other different classifications of MCI (Matthews 2008).

A diagnosis of MCI reputedly allows testing of preventive interventions that would decelerate or slow the progression of MCI to dementia. If the progression of MCI to dementia could be deferred by five years, the prevalence of dementia would decrease by 43% in 2050 (Alzheimer's Association 2010). MCI has an annual progression rate to ADD from 5% to 15%. However, not every person with MCI develops dementia, and a significant number of people recover or stabilise. Therefore, future research should try to clarify which people with MCI develop dementia in order to be able to focus specifically on people who are at high risk of developing dementia. This may possibly explain the failure of therapy to alter the progression to dementia in people with MCI. Other aspects that may contribute to this failure are the disparity in diagnostic criteria and different settings of the studied participants: community, primary, secondary, and research centres (Bruscoli 2004; Mattsson 2009; Petersen 1999; Petersen 2009).

The definition of Alzheimer's disease pathology is over 100 years old. This pathology includes neuritic plaques that contain deposits of amyloid beta (Aβ) and neurofibrillary tangles (Goedert 2006). This pathology is present in approximately 84% of all dementia people (Schneider 2007). Furthermore, Alzheimer's disease pathology is found in 88% of people diagnosed with probable ADD (Schneider 2009). Despite this, Alzheimer's disease pathology may be found concomitantly at autopsy in people thought to have other forms of dementia, such as vascular dementia, Lewy body dementia, or frontotemporal dementia (FTD) (Jellinger 2006). Furthermore, at least five common pathologies have been found in the brains of people who died and were thought to have ADD prior to death (White 2009). Also, Alzheimer's disease pathology was found in 42% of community-dwelling older people without dementia (Schneider 2007). This has generated controversy about the importance of the presence of Alzheimer's disease pathology. The pathology can be associated with aging per se, and for older people the relationship between amyloid plaque burden and cognitive impairment diminishes as age progresses (Savva 2009). Thus this pathology could be an epiphenomenon associated with the presence of dementia, e.g. a by-product of repair mechanisms by vascular damage (de la Torre 2004; Garcia-Alloza 2011).

More recently, the development of Aβ pathology biomarkers in vivo has been suggested as an important advance as a diagnostic tool in the field of Alzheimer's disease, and has promoted the creation of new diagnostic criteria for people without symptoms (preclinical stages), people with MCI, and people with ADD, based on the presence of biomarkers of Alzheimer's disease. These have included Aβ tracers by positron emission tomography (PET) (Albert 2011; Dubois 2014; McKhann 2011; Sperling 2011). However, uncertainties regarding the usability of biomarkers in the diagnosis of dementia still exist, mainly due to variation between biomarker types, criteria for positivity, and differences in methodology (Noel-Storr 2013). This prompted formation of an important initiative called the Standards for Reporting of Diagnostic Accuracy Studies in dementia studies (STARDdem) statement (Noel-Storr 2014). Consequently, clinical properties of dementia biomarkers should not be assumed, and formal systematic evaluations of sensitivity, specificity, and other properties of biomarkers should be performed (Davis 2013).

PET is an imaging technique using compounds labelled with short-lived positron-emitting radionuclides. The use of Aβ ligands permits the in vivo detection of amyloid deposition in the brain. ¹⁸F PET ligands, such as ¹⁸F-Florbetapir and ¹⁸F-Florbetaben, are stilbene derivatives and demonstrate a high binding affinity to A β aggregates. ¹⁸F-Florbetapir has excellent uptake by brain tissue and washout kinetics in mice and monkeys (Choi 2009) and in vitro binding of Aβ plaques in postmortem ADD brain samples (Choi 2009; Lin 2010). In 2010 it was evaluated for the first time in people with ADD and healthy people without ADD”. (Lin 2010; Wong 2010). Similarly, ¹⁸F-Florbetaben, which was first described 10 years ago, is characterised by a high affinity for Aβ (Zhang 2005) and could more accurately diagnose people with ADD (Villemagne 2011).

Another ¹⁸F PET ligand is Flutemetamol. It is a fluorinated tracer, derived from the Pittsburgh Compound B (the first tracer developed). It is characterised by a longer median life of 110 minutes, and has a high affinity for Aβ. Flutemetamol functioning was tested in vivo with healthy people, people with ADD (Nelissen 2009), and in people with MCI (Vandenberghe 2010). Flutemetamol could eventually be used to differentiate between different dementia types, specifically between FTD and ADD, like the other fluorinated tracers ¹⁸F-Florbetaben (Villemagne 2011) and ¹⁸F-Florbetapir (Kobylecki 2015).

The Food and Drug Administration (FDA) and the European Medicines Agency (EMA) approved these three ¹⁸F PET ligands for Aβ. These agencies have stated that a negative scan indicates sparse or no plaques, which is inconsistent with a diagnosis of ADD, thus effectively excluding this diagnosis. A positive ¹⁸F PET ligands scan indicates moderate to frequent or pronounced amyloid neuritic plaques. However, this might also occur in people with other neurological conditions (e.g. Lewy body dementia, Parkinson’s disease dementia) and in older adults with normal cognition. Therefore, a positive result of an ¹⁸F PET ligand scan does not establish the diagnosis of ADD or any other cognitive disorder definitely, and it should be combined with other diagnostic evaluations or instruments. Additionally, the effectiveness and safety of the tests have not been established by predicting development of dementia or other neurological conditions, or by monitoring responses to therapies (EMA 2013; EMA 2014a; EMA 2014b; FDA 2013; FDA 2014a; FDA 2014b).

Despite not being approved for this purpose by the regulatory agencies, research has been conducted in people with MCI to determine whether biomarkers, such as ¹⁸F PET ligands for Aβ, increase the risk of developing dementia over time. The evidence for this is uncertain. For this and other reasons, the National Institute on Aging-Alzheimer's Association (NIA-AA) in the USA established two different criteria for MCI. Firstly, they established the Core Clinical Criteria for use in all clinical settings, without use of biomarkers, and characterised by concerns regarding a change in cognition with impairment in one or more cognitive domains with preservation of independence in functional abilities, therefore no dementia. Secondly, they established the Clinical Research Criteria, which incorporate the use of biomarkers, such as PET amyloid scans, intended for use exclusively in research settings, including academic centres and clinical trials. This will help determine whether positive scans increase the likelihood of progression from MCI to clinical dementia (Albert 2011). Lastly, it is hoped that people with MCI and positive scans will "enrich" clinical trials, and more people who will progress to dementia in a shorter time will be included to allow more efficient studies of treatments and prevention strategies of ADD (CMS 2013).

It is an assumption for some researchers and, in fact, one assumption on which this Cochrane review is predicated, that if a person has both MCI and the pathology of Alzheimer's disease and develops clinical ADD subsequently, then the cause of the initial MCI and of the ADD was the Alzheimer’s pathology. Our approach is an example of assessing diagnostic test accuracy (DTA) using delayed verification of diagnosis. Instead of the reference standard being based on pathology, it is based on a clinical standard and the progression from MCI to ADD or any other form of non-ADD or any dementia. Although, for the reasons stated above, a degree of unreliability has been introduced, defining progression has the advantage of being based on what matters most to people with MCI, their carers, and clinicians involved in their care.

This is a common Cochrane protocol that we will apply to the conduct three separate systematic reviews to assess the DTA of ¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol Aβ binding in the brain and progression of the following:

• From MCI to ADD.
  
• From MCI to any other form of non-ADD.
  
• From MCI to any form of dementia
  

These three Cochrane reviews will assess the following three target conditions.

• ADD (progression from MCI to ADD).
  
• Any other form of dementia (progression from MCI to any other form of non-ADD).
  
• Any form of dementia (progression from MCI to any form of dementia).
  

We will compare the index test results obtained at baseline with the results of the reference standards obtained at follow-up (delayed verification).

The ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) scans are the index tests for the detection of Aβ deposition in the brain region of interest (ROI). The ROI is a selected brain area that physicians create for further study in various anatomical areas of the brain. ¹⁸F PET ligands are molecular biomarkers, and are described as follows.

• ¹⁸F-Florbetapir: (E)-4-(2-(6-(2-(2-(2[¹⁸F]fluoroethoxy)ethoxy)ethoxy)pyridine-3-yl)vinyl)-N-methylbenzamine and also referred to as ¹⁸F-AV-45 (Choi 2009).
  
• ¹⁸F-Florbetaben:¹⁸F-Florbetaben Aβ is a molecular biomarker, ([¹⁸F]BAY 94-9172, trans-4-(N-methyl-amino)-4’-2-[2-(2-[¹⁸F]fluoro-ethoxy)-ethoxy]-ethoxy-stilbene), and is also referred to as BAY 94-9172 or ZK 6013443, which is a polyethylene glycol stilbene derivative (Zhang 2005).
  
• ¹⁸F-Flutemetamol:¹⁸F-Flutemetamol Aβ is a molecular biomarker, described as 6-benzothiazolol, 2-[3-[¹⁸F]fluoro-4-(methylamino)phenyl], and is also referred to as ¹⁸F-3'-F-6-OH-BTA1, ¹⁸F-GE067, AH110690 (Koole 2009; Nelissen 2009).
  

Image Interpretation

Both the FDA and EMA have described the criteria for ¹⁸F PET ligands (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) for Aβ positivity (EMA 2013; EMA 2014a; EMA 2014b; FDA 2013; FDA 2014a; FDA 2014b).

¹⁸ F-Florbetapir:

¹⁸F-Florbetapir diagnosis is by PET image assessment and is designated as either positive or negative by comparison of the radioactivity in cortical grey matter with activity in the adjacent white matter. This determination is made only in the cerebral cortex; the signal uptake in the cerebellum does not contribute to the scan interpretation (e.g. a positive scan may show retained cerebellar grey-white contrast even when the cortical grey-white contrast is lost). Specifically, a positive scan will exhibit one of the following.

• Two or more brain areas (each larger than a single cortical gyrus) in which there is reduced or absent grey-white contrast. This is the most common appearance of a positive scan.
  
• One or more areas in which grey matter radioactivity is intense and clearly exceeds radioactivity in adjacent white matter.
  

¹⁸ F-Florbetaben:

¹⁸F-Florbetaben diagnosis is by PET image assessment, and is defined as positive if the analysis shows the following.

• Moderate or smaller area(s) of tracer uptake equal to or higher than that are presented in the white matter: extending beyond the white matter rim to the outer cortical margin involving the majority of the slices within the respective region.
  
• Pronounced Aβ deposition (a large confluent area of tracer uptake equal to or higher than that presented in white matter extending beyond the white matter rim to the outer cortical margin and involving the entire region including the majority of slices within the respective region) in the grey matter of the following four brain regions: the temporal lobes, the frontal lobes, the posterior cingulate cortex/precuneus, and the parietal lobes.
  

¹⁸ F-Flutemetamol:

¹⁸F-Flutemetamol diagnosis is by PET image assessment, and is defined as positive if analysis shows the following.

• At least one cortical region (Frontal lobes, posterior cingulate and precuneus, lateral temporal lobes, inferolateral parietal lobes, striatum) with reduction or loss of the normally distinct grey-white matter contrast. These scans have one or more regions with increased cortical grey matter signal (above 50% to 60% peak intensity) or reduced (or absent) grey-white matter contrast (white matter sulcal pattern is less distinct), or both.
  
• A positive scan may have one or more regions in which grey matter radioactivity is as intense or exceeds the intensity in adjacent white matter.
  

Readers trained in PET images with the ligands of ¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol, should interpret the Aβ PET images made with those ligands (EMA 2013; EMA 2014a; EMA 2014b; FDA 2013; FDA 2014a; FDA 2014b).

Before the FDA and EMA described the criteria for ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) positivity, the diagnosis of dementia was made using different thresholds. Therefore, it is plausible that we will find studies with differing ¹⁸F PET ligands for Aβ thresholds. Therefore, we plan to use the FDA or EMA criteria applied in each included study to classify participants as either test-positive or test-negative, or alternatively if the ¹⁸F PET ligands for Aβ uptake and retention exceed a certain threshold.

We will consider the measurement of the ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) retention (retention ratio): distribution volume ratio (DVR), standardised uptake value ratio (SUVR), or other ratios. DVR refers to the ratio of the ¹⁸F PET ligands distribution volume in the selected area (ROI) to the distribution volume in the reference area. SUVR is the measured activity in the ROI, which is normalised for body weight, surface area, and injected dose.

The unit of analysis of our Cochrane reviews is the person. It is likely that we will encounter studies that analyse multiple ROIs per person. However, we will only include the pooled results of the ROI in our reviews.

Image analysis: not prespecified (e.g. Statistical Parametric Mapping (SPM) or other image analysis techniques).

Administration Instructions and Recommended Dosing

• Time between ¹⁸F PET ligands for Aβ injection and PET acquisition:
  
  • ¹⁸F-Florbetapir PET images should be acquired in 10 minutes starting from 30 to 50 minutes after intravenous administration (EMA 2013; FDA 2013);
    
  • ¹⁸F-Florbetaben PET images should be acquired in 15 to 20 minutes starting from 45 to 130 minutes after intravenous administration (FDA 2014a) or acquired in 20 minutes starting from 90 minutes after intravenous administration (EMA 2014a);
    
  • ¹⁸F-Flutemetamol PET images should be acquired in 20 minutes starting from 90 minutes after intravenous administration (EMA 2014b; FDA 2014b).
    
• ¹⁸F PET ligands for Aβ injection dose:
  
  • the recommended dose for ¹⁸F-Florbetapir Aβ PET is 370 MBq (10 mCi) as a single intravenous bolus in a total volume of 10 mL or less (EMA 2013; FDA 2013);
    
  • the recommended dose for ¹⁸F-Florbetaben Aβ PET is 300 MBq (8.1 mCi), maximum 30 mcg mass dose (FDA 2014a) or 300 MBq (240 to 360 MBq) administered as a single slow intravenous bolus (6 sec/mL) in a total volume of up to 10 mL (EMA 2014a);
    
  • the recommended dose for ¹⁸F-Flutemetamol Aβ PET is 185 MBq (5.0 mCi) administered as a single slow intravenous bolus (EMA 2014b; FDA 2014b).
    

Although it is inevitable that included studies will have used different imaging protocols, readers' expertise, and varied parameters, the amyloid PET data in these included studies should be technically adequate and acquired at a fully qualified and certified facility.

At this time, the clinical evaluation often has similarities between different countries (Cordella 2013; NICE 2006). It often starts with people experiencing memory complaints detected by themselves or their relatives. Frequently general practitioners or family physicians are consulted, and they often conduct a medical evaluation using a screening test for cognitive impairment. Whenever this screening test is positive, they complete an assessment with a clinical evaluation conducted with laboratory studies that can rule out a secondary cause of cognitive impairment (e.g. hypothyroidism, renal failure, liver failure, vitamin B12 or folate deficiency, and others). In addition, these people are then referred to medical specialists in cognitive disorders (preferably a geriatrician, psychiatrist, or neurologist) in a secondary centre or directly to memory clinics where further clinical assessment, laboratory studies, and cerebral image studies are conducted to confirm the dementia diagnosis.

People with dementia, or their relatives, often directly consult these specialists or specialised memory clinics in the study of cognitive disorders. Therefore, the performed diagnostic tests will probably vary according as to whether it is a primary consultation or already referred from primary to specialist care, or if the people have different clinical stages of the disease (MCI, mild, moderate, or severe dementia). Due to these differing pathways, the use of ¹⁸F PET ligands for Aβ will be mainly used in specialist consultations and memory clinics as an addition to clinical evaluation or other tests, helping in a clinical setting to discard a diagnosis of Alzheimer's dementia with a negative scan in a person with clinical dementia and doubts about the ethology (e.g. FTD versus ADD). Otherwise, it might be used solely in the research field in people with an MCI for the enrichment of clinical trials with people with MCI with positive PET scan to study preventive interventions or if the results of PET Aβ imaging lead to improved health outcome.

Alternative test(s)

Currently there are no standard practice tests available for the diagnosis of dementia. Below, we have listed the alternative tests that we have excluded from this Cochrane review. The Cochrane Dementia and Cognitive Improvement Group is in the process of conducting a series of DTA reviews of biomarkers and scales (see list below).

• ¹⁸F-FDG-PET (PET F-fluorodeoxyglucose) (Smailagic 2015).
  
• 11C-PIB-PET (PET-Pittsburgh compound B) (Zhang 2014).
  
• Cerebrospinal fluid (CSF) analysis of Aβ and tau (Kokkinou 2014; Ritchie 2013; Ritchie 2014).
  
• Structural magnetic resonance imaging (sMRI) (Filippini 2012).
  
• Neuropsychological tests (Mini-Mental State Examination (MMSE); MiniCOG; Montreal Cognitive Assessment (MoCA) (Arevalo-Rodriguez 2015; Chan 2014; Creavin 2016; Davis 2015; Fage 2015; Seitz 2014).
  
• Informant interviews (Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE); AD8) (Harrison 2014; Hendry 2014; Lees 2014; Harrison 2015; Quinn 2014).
  
• APOE-ε4 (Elias-Sonnenschein 2014a; Elias-Sonnenschein 2014b; Elias-Sonnenschein 2014c).
  
• Single-photon emission computed tomography (SPECT) brain imaging (Archer 2015; McCleery 2015).
  

Accurate and early diagnosis of Alzheimer's disease is crucial for planning healthcare systems, because the costs of dementia are currently at least 1% of the world's GNP (WHO 2012).

¹⁸F PET ligands for Aβ biomarkers are approved for use in the clinical field mainly in people who are diagnosed clinically with dementia of uncertain aetiology, in which case diagnosis of ADD can be discarded if the test is negative. Even though ¹⁸F PET ligands for amyloid biomarkers are not approved for this purpose, these biomarker tests are currently being used in the research field to search for the accurate identification of people with MCI who would progress to Alzheimer’s disease or other forms of dementia. Amyloid β tracers by PET have been included in newly diagnostic criteria in the study in people with MCI (Albert 2011; Dubois 2014). However, some uncertainties exist about DTA performance in clinical settings, especially in older people (Richard 2012).

It is currently believed that if the health system can identify which people are at high-risk of progressing from MCI to dementia, it can focus on improving opportunities for appropriate contingency planning for them. Proper recognition of the disease may also help prevent inappropriate and potentially harmful admissions to hospital or institutional care (NAO 2007), and enable the development of new treatments designed to delay or prevent progression to more debilitating stages of the disease. Additionally, this may demonstrate a real clinical benefit for people and caregivers, and will reduce health system costs.

These three separate Cochrane reviews will assess DTA with ¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol Aβ PET in people with MCI.

To determine the diagnostic test accuracy (DTA) of the ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) as the index tests for detecting participants with mild cognitive impairment (MCI) at baseline who would clinically progress to Alzheimer's disease dementia (ADD), or other forms of non-ADD, or any form of dementia at follow-up.

To investigate the heterogeneity of the DTA in the included studies, we will evaluate the spectrum of people, referral centres, clinical criteria of MCI, ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) techniques, reference standards used, duration of follow-up, aspects of study quality, and conflicts of interest.

Types of studies

We will include longitudinal studies that have prospectively defined cohorts with any accepted definition of mild cognitive impairment (MCI) with baseline ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) scans and the reference standard (see Index tests and Reference standards below). We will obtain the results at the follow-up of the studies. These studies necessarily will employ delayed verification of progression to dementia and are sometimes labelled as 'delayed verification cross sectional studies' (Bossuyt 2008; Knottnerus 2002). We will include case control studies when they incorporate a delayed verification design. This will occur in the context of a cohort study, so these studies are invariably diagnostic-nested case-control studies.

Participants

Participants recruited and clinically classified as those with MCI at baseline will be eligible for inclusion. We will establish the diagnosis of MCI using the Petersen criteria or revised Petersen criteria (Petersen 1999; Petersen 2004; Winblad 2004), the Matthews criteria (Matthews 2008), CDR = 0.5 (CDR structured interviews collects information from both the collateral source and the subject regarding memory, orientation, judgment and problem solving, community affairs, home and hobbies, and personal care, where the range of possible scores varies from none=0 point to severe=3 points) (Morris 1993), the National Institute on Aging-Alzheimer's Association (NIA-AA) core clinical criteria (Albert 2011), or a combination.

We will exclude studies that included people with MCI possibly caused by any of the following.

• Current or a history of alcohol or drug abuse.
  
• Central nervous system (CNS) trauma (e.g. subdural hematoma), tumour, or infection.
  
• Other neurological conditions (e.g. Parkinson’s or Huntington’s diseases).
  

Index tests

The index test in these three Cochrane reviews will be ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) biomarker tests. We will use the criteria and cut-off values for test positivity as reported in the included studies. We will consider positivity for¹⁸F PET ligands for Aβ scans (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) uptake and retention exceeding a certain threshold.

Target conditions

There are three target conditions in these three Cochrane reviews.

• Alzheimer’s disease dementia (ADD) (progression from MCI to ADD).
  
• Any other forms of dementia (progression from MCI to any other forms of non-ADD).
  
• Any form of dementia (progression from MCI to any form of dementia).
  

Reference standards

The reference standard will be the progression to the target conditions evaluated by a physician with expertise in the dementia field (preferably a geriatrician, psychiatrist, or neurologist). For the purpose of these three Cochrane reviews, we will accept several definitions of ADD. We will include studies that applied the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association (NINCDSADRDA) criteria (McKhann 1984), the Diagnostic and Statistical Manual of Mental Disorders (DSM) criteria (APA 1987; APA 1994), and the International Classification of Diseases (ICD) (ICD-10) criteria for ADD. Notably, different iterations of these standards may not be directly comparable over time (e.g. APA 1987 versus APA 1994). Moreover, the validity of the diagnoses may vary with the degree or manner in which the criteria have been operationalised (e.g. individual clinician versus algorithm versus consensus determination). We will consider all these issues when we interpret the results, and will use sensitivity analyses when appropriate.

Similarly, we will accept differing clinical definitions of other dementias. For Lewy Body Dementia the reference standard is the McKeith criteria (McKeith 1996; McKeith 2005); for Fronto-Temporal Dementia the Lund criteria (Boxer 2005; Brun 1994; Neary 1998), the DSM criteria (APA 1987; APA 1994), the ICD criteria (ICD-10), or the International Behavioural Variant FTD Criteria Consortium (Rascovsky 2011); and for vascular dementia the National Institute of Neurological Disorders and Stroke and Association Internationale pour la Recherché et l'Enseignement en Neurosciences (NINDS-AIREN) criteria (Román 1993), the DSM criteria (APA 1987; APA 1994), or the ICD criteria (ICD-10).

The time interval over which the progression from MCI to ADD (or other forms of dementia) occurs is very important. We will use one year as the minimum period of delay in the verification of the diagnosis (the time between the assessment at which a diagnosis of MCI is made and the assessment at which the diagnosis of dementia is made).

Electronic searches

We will search MEDLINE (Ovid SP) from 1946 to present; EMBASE (Ovid SP) from 1974 to present ; PsycINFO (Ovid SP) from 1806 to present ; BIOSIS Citation Index (Thomson Reuters Web of Knowledge) from 1922 to present; Web of Science Core Collection, including the Science Citation Index (ISI Web of Knowledge) and the Conference Proceedings Citation Index (Thomson Reuters Web of Knowledge) from 1946 to present; LILACS (Bireme); CINAHL (EbscoHOST) from 1980 to present; ClinicalTrials.gov (https://clinicaltrials.gov); and the World Health Organization International Clinical Trials Registry Platform (WHO ICTRP) (http://www.who.int/ictrp/search/en/). We will also search ALOIS (Cochrane Register of Studies Software).

See Appendix 1 for details of the sources and search strategies that we will use. We will not apply any language restrictions to the electronic searches. The searches are entirely index test focused for maximum sensitivity and we will not apply any methodological filters.

Searching other resources

We will examine the reference lists of all relevant studies for additional studies. We will also search the Database of Abstracts of Reviews of Effects (DARE) via the Cochrane Library: www.cochranelibrary.com), the HTA Database (via the Cochrane Library: www.cochranelibrary.com), the Aggressive Research Intelligence Facility (ARIF) database (www.arif.bham.ac.uk) for other related systematic diagnostic accuracy reviews, and the International Federation of Clinical Chemistry and Laboratory Medicine Committee for Evidence-based Laboratory Medicine database (C-EBLM) (http://www.ifcc.org/ifcc-education-division/emd-committees/c-eblm/evidence-based-laboratory-medicine-c-eblm-base).

We will check the reference lists of any relevant studies and systematic reviews, and perform citation tracking using Science Citation Index to identify any relevant studies.

Selection of studies

Two review authors (GM, RV) will independently screen the retrieved titles and abstracts for potentially eligible studies. A third review author (PF) will resolve any disagreements between the two review authors. The two review authors (GM, RV) will then independently assess the full-text articles of the selected studies against the inclusion criteria. They will resolve any disagreements through discussion or, where necessary, will consult a third review author (PF) who will act as an arbitrator. When a study does not present all relevant data for creating 2×2 table, we will contact the study authors directly to request further information. When more than one article presents data on the same population, we will include the primary article, which will be the article with the largest number of people or with the most informative data (e.g. longest time of follow-up in the primary outcome).

Data extraction and management

We plan to extract the following data regarding the study characteristics.

• Bibliographic details of primary paper:
  
  • author, title of study, year, and journal.
    
• Basic clinical and demographic details:
  
  • number of participants;
    
  • clinical diagnosis;
    
  • MCI clinical criteria;
    
  • age;
    
  • gender;
    
  • sources of referral;
    
  • participant recruitment;
    
  • sampling procedures.
    
• Details of the index test:
  
  • method of the ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) tests administration, including those who administered the test;
    
  • thresholds used to define positive and negative tests;
    
  • other technical aspects as seemed relevant to the review, e.g. brain areas.
    
• Details of the reference standard:
  
  • definition of ADD and other dementias used in reference standard;
    
  • duration of follow-up from time of index test performed to defining ADD and other dementias by reference standard; one year to less than years; two years to less than four years; and more than four years. If participants have been followed for varied amounts of time we will record a mean follow-up period for each included study. If possible, we will group those data into minimum, maximum, and median follow-up periods, which may then become the subject of subgroup analyses;
    
  • prevalence or proportion of population developing ADD and other dementias, with severity if described.
    

We will create 2×2 tables (cross-relating index test results of the reference standards) as shown in Appendix 2. For each included study, we will record the number of people lost to follow-up. We will also extract data necessary for the assessment of quality as defined below. Two review authors (GM, RV) will independently perform data extraction. We will resolve any disagreements regarding data extraction by discussion, or consult a third review author (PF) if necessary.

Assessment of methodological quality

We will assess the methodological quality of each included study using the Quality Assessment of Diagnostic Accuracy Studies 2 tool (QUADAS-2) (Whiting 2011), as recommended by Cochrane. This tool is comprised of four domains: patient selection, index test, reference standard, and patient flow.

Two review authors (GM, RV), who will be blinded to each other’s scores, will independently perform the QUADAS-2 assessment. They will resolve any disagreements by discussion or, if necessary, consult a third review author (PF) who will act as an arbitrator. We will assess each domain in terms of risk of bias, and will also consider the first three domains in terms of applicability concerns. In Appendix 3 we have detailed the components of each of these domains and provided a rubric that shows how we will make judgements concerning risk of bias. Key areas important to quality assessment are participant selection, blinding, and missing data.

We will include three additional signalling questions on our checklist.

• Was the PET scan interpretation done by a trained reader physician? (We will included this under the ’Index test’ domain.)
  
• Was there a clear definition of a positive result? (We will included this under the ’Index test’ domain.)
  
• Was the study free of commercial funding? (We will included this under the ’flow and timing’ domain.)
  

We will include the item pertaining to the PET scan interpretation and the definition of positive results to take into account the subjective nature of ¹⁸F- PET scan image interpretation, which may be based on a variety of different criteria, such as extensive clinical experience, different standard uptake values (SUV), different morphological features, or a combination of the aforementioned. We will include the third additional item in order to record any potential bias resulting from commercial interest in the results.

We will not use QUADAS-2 data to form a summary quality score. We will produce a narrative summary that describes the numbers of included studies that are at high, low, or unclear risk of bias as well as concerns regarding applicability, which we have described in Appendix 4.

Statistical analysis and data synthesis

We will apply the DTA framework for the analysis of a single test and will extract the data from each included study into a 2×2 table, we will show the binary test results cross-classified with the binary reference standard, and we will ignore any censoring that might have occurred. We acknowledge that such a reduction in the data may represent a significant oversimplification. We will therefore adopt an intention-to-diagnose (ITD) approach as well. If possible, we will present what the result would be if all dropouts (Individuals who leave the study prior to completion of follow up) would have developed dementia, and if all dropouts would not have developed dementia.

We will use data from the 2×2 tables abstracted from the included studies: true positive (TP), false negative (FN), false positive (FP), true negative (TN) and entered into Review Manager (RevMan) (Review Manager 2014) to calculate the sensitivities, specificities, and their 95% confidence intervals. We will also present individual study results graphically by plotting estimates of sensitivities and specificities in both a forest plot and a receiver operating characteristic (ROC) space. If an individual included study publishes more than one threshold, we will present the graphical findings for all reported thresholds. However we will avoid inclusion of study data in the calculation of a summary statistic on more than one occasion (in the same setting) by using only the threshold, which is considered to be ’standard practice’ for the target population in question. We will not pool studies across settings. If we are unable to agree upon a standard practice for the index test and the target population in question, we will use the optimal threshold (i.e. the threshold nearest to the upper left corner of the ROC curve) in calculating the summary ROC curve in RevMan (Review Manager 2014) and for any subsequent meta-analyses; we recognise that this may lead to an overestimation of diagnostic accuracy (Leeflang 2008). If there are common thresholds across included studies we will also consider the bivariate random-effects approach (Reitsma 2005).

If there is sufficient data we will meta-analyse the pairs of sensitivity and specificity. The preferred approach would be the hierarchical summary ROC curve (HSROC) method proposed by Macaskill 2010 and Rutter 2001 because implicit thresholds are expected in primary studies. We will conduct these analyses using the Statistical Analysis Software (SAS) (SAS Institute 2011).

We plan to segment analyses into separate follow-up mean periods for the delay in verification: one year to less than two years; two to less than four years; and greater than four years. In this we plan to clearly note where the same included studies contributed to the analysis for more than one reference standard follow-up interval.

We will explore the implications of any summary accuracy estimates not affected by heterogeneity emerging by considering the numbers of FP and FNs in populations with different prevalence of MCI, and by presenting the results as natural frequencies and using alternative metrics such as likelihood ratios and predictive values.

We will prepare a summary of findings table irrespectively.

Investigations of heterogeneity

We plan to investigate the effects of the following factors.

• Spectrum of people (mean age, gender, Mini-Mental State Examination (MMSE) score, APOE ε4 status). For age, we will separately examine any studies that included 30% of people below the age of 65.
  
• Referral centres: primary care, memory clinics, and hospitals.
  
• Clinical criteria of MCI: Petersen criteria, revised Petersen criteria, CDR = 0.5 criteria, and different MCI classification (Matthews 2008).
  
• Index test: thresholds, if stated; differences in ¹⁸F-PET ligands for Aβ retention ratio; differences in image analysis; differences in time between ¹⁸F-PET ligands for Aβ and PET acquisition; differences in¹⁸F-PET ligands for Aβ injection dose; differences in ¹⁸F-PET ligands for Aβ retention detecting regions.
  
• Reference standard(s) used: NINCDS-ADRDA, DSM, or ICD-10 for ADD.
  
• Duration of follow-up: one year to less than two years; two to less than four years; and greater than four years.
  
• Aspects of study quality, particularly inadequate blinding, and loss to follow-up: we will consider separately those studies that have more than 20% drop-outs.
  
• Conflicts of interest.
  

In preliminary analyses, we will examine forest plots of sensitivity and specificity, and summary ROC plots to explore the effect of each of these factors. If there will be sufficient studies, we plan to perform a meta-regression by including each potential source of heterogeneity as a covariate in the bivariate or in the HSROC model.

Sensitivity analyses

We will investigate the influence of study quality on overall DTA of the¹⁸F-PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) index, using QUADAS-2. In a sensitivity analysis we will exclude studies at high risk of bias in any domain or high concern applicability as assessed by QUADAS-2, and we will compare the results with those from the primary analysis.

We will ignore any losses to follow-up in the primary analyses. Such a reduction in the data might represent a significant oversimplification. Therefore, we will adopt an ITD approach in sensitivity analyses wherein we will impute missing data under these following assumptions.

• All dropouts developed dementia.
  
• All dropouts did not develop dementia.
  

Differing length of follow-up may affect outcomes. We have set a minimum mean time to follow-up assessment of 12 months. We will investigate the stability of results when we exclude studies with a follow-up period of between one year to less than two years from the analyses.

When the criteria used for the clinical diagnosis of dementia are not among the acceptable reference standards of this Cochrane review, we will likewise exclude these studies in a sensitivity analysis to test whether the results differ from the analysis including all studies.

Assessment of reporting bias

We will not investigate reporting bias because of current uncertainty about how it operates in DTA studies and the interpretation of existing analytical tools, such as funnel plots (Leeflang 2008).

Gabriel Martínez is a PhD candidate in Methodology of Biomedical Research and Public Health at the Department of Paediatrics, Obstetrics and Gynaecology and Preventive Medicine, Universitat Autònoma de Barcelona, Barcelona, Spain.

We thank Anna Noel-Storr, Information Specialist of the Cochrane Dementia and Cognitive Improvement Group, for her assistance with the design of the search strategy.

Table 1. Progression from mild cognitive impairment (MCI) to Alzheimer’s disease dementia (ADD)

ADD: Alzheimer's disease dementia TP: true positive FP: false positive FN: false negative TN: true negative

Table 2. Progression from mild cognitive impairment (MCI) to non-Alzheimer’s disease dementia (ADD)

ADD: Alzheimer's disease dementia MCI: mild cognitive imapirment

TP: true positive FP: false positive FN: false negative TN: true negative

Table 3. Progression from mild cognitive impairment (MCI) to any form of dementia

TP: true positive FP: false positive FN: false negative TN: true negative

Table 4. Review question and inclusion criteria

MCI: mild cognitive impairment ADD: Alzheimer's disease dementia NINCDS-ADRDA: National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association DSM: Diagnostic and Statistical Manual of Mental Disorders

ICD: International Classification of Diseases NINDS-AIREN: National Institute of Neurological Disorders and Stroke and Association Internationale pour la Recherché et l'Enseignement en Neurosciences FTD: Frontotemporal dementia

Anchoring statements for quality assessment of ¹⁸F PET ligands for Aβ diagnostic studies

We have provided some core anchoring statements for quality assessment in the diagnostic test accuracy (DTA) review of the ¹⁸F PET ligands for Aβ (¹⁸F-Florbetapir, ¹⁸F-Florbetaben, and ¹⁸F-Flutemetamol) biomarkers in dementia. These statements are designed for use with the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool and are based on the guidance for quality assessment of DTA reviews of Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) in dementia (Quinn 2014). In assessing individual items, the score of unclear should only be given if there is genuine uncertainty. In these situations we will contact the relevant study teams for additional information. Whenever we score one question as high risk of bias, we will consider the study having a high risk of bias.

Table 5. Anchoring statements to assist with the 'Risk of bias' assessment

DSM: Diagnostic and Statistical Manual of Mental Disorders ICD: International Classification of Diseases NINCDS-ADRDA: National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association FTD: Frontotemporal dementia NINDS-AIREN: National Institute of Neurological Disorders and Stroke and Association Internationale pour la Recherché et l'Enseignement en Neurosciences MCI: mild cognitive impairment

Gabriel Martínez and Leon Flicker conceived, designed and drafted the protocol.

Robin WM Vernooij, Paulina Fuentes Padilla, Javier Zamora, Marta Roqué i Figuls, Gerard Urrútia, and Xavier Bonfill Cosp designed and drafted the protocol.

Gabriel Martínez has no known conflicts of interest.
Leon Flicker has no known conflicts of interest.
Robin WM Vernooij has no known conflicts of interest.
Paulina Fuentes Padilla has no known conflicts of interest.
Javier Zamora has no known conflicts of interest.
Marta Roqué i Figuls has no known conflicts of interest.
Gerard Urrútia has no known conflicts of interest.
Xavier Bonfill Cosp has no known conflicts of interest.

• Iberoamerican Cochrane Centre, Barcelona, Spain.
  
  Logistic
  

• National Institute for Health Research (NIHR), UK.
  
  This protocol was supported by the NIHR, via Cochrane Infrastructure funding to the Cochrane Dementia and Cognitive Improvement group. The views and opinions expressed therein are those of the protocol authors and do not necessarily reflect those of the Systematic Reviews Programme, the NIHR, the NHS, or the Department of Health