Introduction
Here, we introduce a new clinical multi-centre study (Memory Clinic: MemClin) intended to fill the gap for large-scale clinical studies aimed at describing early stages of cognitive decline regarding diagnostic accuracy, longitudinal prognosis, differential diagnostics, and at creating better harmonised examination protocols.
The risk of dementia rises sharply with age [
1], and according to the United Nations the aging population is increasing [
2]. Dementia affects approximately 5% of the world’s elderly population [
1], and the percentage of people with Alzheimer’s disease (AD) increases exponentially: 3% between 65 and 74 years, 17% between 75 and 84 years, and 32% older than 85 years [
3]. These statistics describe the magnitude of the burden of cognitive disorders on the health-care system [
3]. Decline in cognitive functions is also a general feature of normal aging reflected by deterioration in cognitive domains such as episodic memory [
4], processing speed, and executive functions [
5,
6]. A great clinical challenge is to accurately distinguish between normal age-related cognitive decline and the earliest signs of dementia. In keeping with that notion, the concept of mild cognitive impairment (MCI) was initially established as a prodrome for persons at elevated risk of developing AD [
7]. MCI has been verified as a risk factor for dementia [
8,
9], but nonetheless some MCI individuals remain stable or recover a normal cognitive status over time [
10]. Recently, criteria for subjective cognitive decline (SCD) have been proposed [
11,
12]. SCD is a concept contingent on self-reported cognitive decline but with unimpaired performance on cognitive test measures. In addition, SCD refers to progressive change from a previous cognitive level and not just an isolated complaint. It has been shown that SCD predicts dementia [
13,
14]. However, MCI and SCD individuals may have a heterogeneous aetiology for cognitive impairment in early phases, with diverse clinical manifestations and different temporal trajectories of both their somatic and cognitive profiles. Thus, the cognitive and physiological trajectories in both SCD and MCI patients’ need to be explored further. In the current article, given the insufficient knowledge about change over time at the baseline examination, the term subjective cognitive impairment (SCI) will be used instead of SCD.
Misdiagnosis of dementia is an important clinical problem affecting the patient’s life and may have profound implications for their close relations. Increased knowledge on how neuropathological features relate to clinical criteria is urgently needed to prevent such outcomes [
15]. To increase knowledge of the pathophysiology in dementias such as AD, biomarkers have been increasingly important [
16,
17]. In keeping with that, an unbiased classification scheme (A/T/N) for AD has been proposed [
18]. We have recently shown that the A/T/N scheme is a promising tool for increasing the degree of certainty in diagnostic and prognostic procedures when used in conjunction with information on brain atrophy (as assessed with magnetic resonance imaging (MRI)) [
19]. However, the A/T/N scheme and other research findings need to be further validated in large-scale community cohorts such as MemClin. Due to potential heterogeneity of consensus procedures, and comorbidities often seen in memory clinics, these are perfect settings for investigating the validity of the A/T/N biomarker scheme regarding non-AD related dementias, injuries and somatic conditions. There are large-scale databases [
20‐
23] on cognitive impairment, in research settings such as the Alzheimer’s disease neuroimaging initiative (ADNI) [
24], in clinical settings such as the Gothenburg MCI study [
25]. In comparison, the MemClin project strength is that it covers 9 out of 10 memory clinics in a city population greater than two million inhabitants. Thus, this is a naturalistic study with an unselected inclusion, and due to the magnitude of MemClin, it will be possible to regularly evaluate different aspects of the more unusual cognitive conditions, and neurodegenerative dementias such as dementia with Lewy Bodies, Parkinson’s disease dementia, subtypes of fronto-temporal dementias, and patients with dementia whose clinical diagnosis cannot be specified (“UNS”). This will contribute important information on these subtypes of dementia pathology. Also including these subtypes will provide a rare ability to generalise findings from MemClin to the general clinical population with cognitive impairment, for instance through improved differential diagnostics. Importantly, MemClin has the ability to follow the diagnostic trajectories of patients with SCI and MCI. Taken together, our ongoing MemClin project has a large-scale magnitude and population-based multi-centre character is envisioned to provide a platform concerning clinically relevant research for many years to come, targeting early clinical manifestations of diverse neurodegenerative disorders and other cognitive conditions. The MemClin research project will address three main objectives:
(I) To examine how cognitive profile, brain imaging [MRI and/or computed tomography (CT)], cerebrospinal fluid (CSF) biomarker candidates, and other health assessments (such as blood sampling and body-mass index) predict the longitudinal progression from SCI, MCI to dementia.
(II) To examine how specific cognitive tests, brain imaging, CSF, and other health assessments and the combination of those, relate to differential diagnostics in various cognitive stages and neurodegenerative disorders.
(III) To examine specific sub-types (e.g., interactions of cardiovascular pathology to AD pathology as well as atrophy patterns) of AD with respect to progression and diagnostic accuracy, but also to increase knowledge of “unspecified dementias (UNS)” (i.e., when a specific clinical diagnosis cannot be established).
The main objective of the current article is to describe the study design, materials and methods, and patient characterisation of the MemClin project. In addition, we will report descriptive data from the ongoing data collection.
Discussion
The main purpose of this article was to introduce the ongoing MemClin project in which we coordinate data collection across nine out of ten memory clinics in the greater Stockholm catchment area in Sweden. We also briefly presented patient characteristics and preliminary descriptive results. At present, the data collection of the MemClin project is in progress and more than 1500 patients are currently included. We expect to be collecting data in full scale throughout 2019–2022, gathering data for approximately 1200–1500 new patients annually. All memory clinics in the greater Stockholm area largely use similar examination methods, which facilitates data collection and aggregation. Establishing a large-scale clinical database of relevant data will enable us to build the robust models required to improve diagnostic dementia accuracy, differential diagnostics, and hopefully shorten the time from initial exam to a reliable diagnosis. Ultimately, the MemClin effort serves the overarching purpose of better and more cost-efficient healthcare for future memory clinics.
The preliminary data herein was in line with the expected outcome. This was for example illustrated as overall superior cognitive performances observed in SCI patients, intermediate cognitive performances in MCI patients, and inferior performances on cognitive measures such as episodic memory scores in patients with AD. Additional analyses will be conducted when the MemClin project can provide increased statistical power and more advanced statistical modelling.
There are large-scale databases available that aim to improve quality of dementia care by evaluating adherence to national guidelines and support care harmonisation across clinics, such as the SveDem national quality registry [
28]. However, those registries do not provide information on extensive neuropsychological measures and early phase cognitive impairment (i.e., SCI and MCI). MemClin will provide this essential information. Furthermore, due to the magnitude of MemClin, it will be possible to regularly evaluate different aspects of the more unusual neurodegenerative dementias such as dementia with Lewy Bodies, Parkinson’s disease dementia, subtypes of fronto-temporal dementias, and patients with dementia whose clinical diagnosis cannot be specified (“UNS”). This will contribute important information n these subtypes of dementia pathology. Also including these subtypes will provide a rare ability to generalise findings from MemClin to the general clinical population with cognitive impairment, for instance through improved differential diagnostics. Most importantly, the project will generate harmonised examination protocols with highest possible sensitivity to prodromal stages of dementias. By harmonising diagnostic procedures across clinics, the MemClin effort serves the overarching purpose of better and more cost-efficient healthcare for future memory clinics. Furthermore, the knowledge generated from MemClin will provide dementia examinations with a higher quality and diagnostic accuracy and thus contribute to a more adequate caretaking and less acute healthcare arrangements. This improvement will have implications on the quality and accuracy of dementia diagnosis within the Stockholm catchment area, and also potentially domestically and internationally. Lastly, because data collection is conducted as part of the clinical routine, new reliable findings can be applied back to the clinics within a relatively short timeframe of implementation.
There are some limitations to the current study. Clinicians at these memory clinics use visual ratings (CT/MRI) from clinical radiologists and thus we introduce unknown variability that could affect inter-rater reliability. Further, at present we are missing values for the visual rating scales of some patients despite having CT/MRI data available. Therefore, we will re-evaluate the ratings with expert neuroradiologists to harmonise the ratings, minimise inter-rater disagreement, and provide comparable atrophy estimates for all MemClin CT/MRI data. We also plan to conduct automated brain-imaging analyses on these images [
38]. Since this is a multi-centre study there will be different practices and priorities at the centres when it comes to cognitive testing, CSF sampling, additional CT/MRI evaluations, and other health measures in relation to the follows. Due to this potential inconsistency between centres, we are planning to follow the patients with SCI within the MemClin project to receive comparable and useful data for those in the earliest phases of cognitive impairments. In addition, those patients that the MemClin project will not be able to follow over time, may still become available for diagnostic follow via the SveDem registry once they develop dementia [
28]. We are aware that potential biases can occur considering incomplete data that effect the validity of the research results. Our approach to follow the diagnostic trajectories in patients with SCI from different sources will decrease the risk for biases. However, we will also conduct sensitivity analyses to assess the effect of dropout and missing data.
Since part of the reported data in the MemClin project are included in the diagnostic criteria, there is a risk of circularity bias. Therefore, circularity bias needs to be considered when examining the diagnostic estimates. Cognitive tests are included in the diagnostic criteria. CSF biomarkers are used for the diagnosis in approximately 40% of the MemClin patients. MRI and/or CT are used in an unstructured way to ascertain brain atrophy and exclude tumours, etc. in approximately 80% of the MemClin patients. Thus, there may be certain risk of circularity bias when addressing our project’s aims I and II, especially for cognitive tests and visual ratings from MRI/CT. Therefore, we will consider this circularity bias when examining the diagnostic estimates, for example by comparing the value of measures at different levels of bias (e.g. cognitive tests vs. CSF) as well as by re-diagnosing patients for specific studies.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.