The present studies have made a number of contributions to the field of AD- related SCD research. Study 1 has highlighted the fact that cognitively unimpaired elderly General Practice patients, a sample highly relevant to the health care sector, who report SCD with associated concerns (SCD+C) are of a 2.5-fold increased risk to develop AD dementia over a period of 6 years compared to individuals with normal cognition and without SCD. Study 1, thus, suggests that SCD in the memory domain might be useful as a predictor of incident AD dementia, even in the absence of overt neuropsychological testing deficits. This finding confirmed earlier findings of the AgeCoDe study with shorter follow-up time (Jessen et al. 2010) and is in line with recent findings from other epidemiologic studies investigating SCD as a risk factor for dementia (Mitchell et al. 2014).
Beyond these confirmatory results, study 1 suggested that the qualitative feature of concerns/worries about the self-perceived memory decline might be of specific predictive value as the risk in the SCD+C group was also significantly increased compared to those individuals with SCD but without associated concerns. Study 1, thus, provided strong evidence for the recent inclusion of concerns regarding experienced cognitive decline, especially in the memory domain, as one of several features that may increase the likelihood of underlying AD in subjects with SCD (Jessen et al. 2014a). As an open question for future research it remains to be determined why the appraisal of self-experienced cognitive decline as worrying is associated with such an elevated risk of AD dementia compared to SCD without concerns. The results suggest that AD- related memory decline is experienced in a different quality (i.e. as more serious and
therefore worrying) compared to memory decline related to other factors such as normal aging. Following this idea, further research should try to determine what specific qualitative aspects of the self-experienced cognitive decline (and what other factors) may determine the emergence of patients’ concerns. Research in this regard will contribute to better operationalization and measurement of AD-related SCD.
Importantly, in study 1 the AD dementia risk in individuals with SCD+C was not substantially different from those individuals whose memory performance fell in the range of early amnestic MCI (risk of AD dementia in this group was also about 2.5-fold increased compared to the cognitively normal control group). It, thus, seems that the transition of pre-MCI SCD to early MCI is fluent and the definition of a strict border between these two stages is arbitrary. This clearly highlights the problems associated with neuropsychological cutoffs to define, and differentiate from one another, the presumed late preclinical stage of AD (clinically manifesting as “only SCD”) and the early MCI stage.
Study 2 focused on the interplay between objective and subjective cognitive decline with regard to risk of AD dementia in a large memory clinic sample and was the first study to directly test a prediction hypothesis derived from the recently proposed working model of the temporal evolution of SCD in the course of AD (Jessen et al.
2014a). From a statistical point of view, the assumptions on the interplay between objective memory performance and SCD outlined above can be modeled by a moderating effect of objective memory performance on the relationship between SCD and incident AD dementia: The predictive value of SCD for incident AD dementia should be highest at subtle memory impairment and should decrease as memory impairment progresses. Results of study 2 showed a significant main effect of SCD+C on incident AD dementia. Across the whole sample MCI patients with memory concerns were of 2.5-fold increased risk to convert to AD dementia compared to those MCI patients without concerns about their experienced memory decline. The hypothesized relationship between objective and subjective cognitive decline with regard to risk of AD dementia could also be demonstrated by a significant moderating effect as proposed above. This finding has also relevance for the widely received dynamic biomarker model for the development of AD (Jack et al. 2013). In this model, biomarkers of AD pathology as well as cognitive and functional performance are modeled as sigmoid-shaped, ascending curves of pathology/impairment. However, the
development of AD-related SCD seems to markedly differ from these trajectories. The SCD trajectory may rather divert from the trajectory of objective impairment in the stage of late MCI/early AD dementia and may be better modeled as an inverse U-shaped function. Thus, the association of SCD with biomarkers and cognitive/functional status is variable throughout the course of AD. Since SCD is gaining even further interest in the scientific field of AD, a revised biomarker model of AD should incorporate the aforementioned diverging trajectory of SCD and support this hypothesized trajectory with empirical, longitudinal data on intra-individual changes in SCD.
Study 3 demonstrated that, in a memory clinic sample, specific aspects of SCD quantity and quality were predictive of a CSF profile indicating AD pathology, i.e.
presence of “prodromal AD” according to the IWG2 (Dubois et al. 2014) and NIA-AA (Albert et al. 2011) criteria. Risk of a prodromal AD CSF profile increased by 73% for a one standard deviation increase in scores on a quantitative SCD scale and was approximately doubled in those MCI patients that reported memory concerns, respectively. These results, thus, highlighted, together with the longitudinal findings of study 2, the usefulness of SCD operationalizations that extend beyond the broad clinical SCD criterion as part of the general MCI diagnostic procedures. This argues for a refined SCD assessment in MCI and might again be especially important in the early stage of MCI.
In practice, clinicians are indeed often confronted with cases that fall on the border between normal test performance and early mild cognitive impairment. The DCN memory clinic cohort, initially drawn from patients that actively sought evaluation at one of the DCN memory clinics, contains a substantial number of such mildly impaired patients. Therefore, not only has the DCN MCI sample a high external validity. It follows from this that the diagnostic entity of SCD without clearly determinable cognitive impairment is highly relevant for memory clinic routine practice. While such patients have long been considered “the worried well” or those with depression related complaints, the recent literature provides growing evidence that this clinical entity of pre-MCI SCD should be taken seriously and better characterized in future studies (Jessen et al. 2014a).
This, however, also implies that the definition of objective impairment at the MCI level, which will serve as an exclusion criterion for studies on pre-MCI SCD,
needs to be sharpened as well as harmonized across those studies. It can be argued here that, for the purpose of distinguishing a group of individuals with pre-MCI SCD from those with MCI, a more robust neuropsychological MCI definition, compared to that applied in studies such as the DCN cohort should be chosen (Edmonds et al. 2014;
Bondi et al. 2014). Recently proposed neuropsychological MCI criteria which balance reliability (i.e. stability of the MCI diagnosis across time) and sensitivity to incipient AD (Jak et al. 2009; Bondi et al. 2014) could be used to define such a robust MCI group from which a pre-MCI SCD group could then be distinguished by not meeting these neuropsychological MCI criteria.
Following the logic from the working model outlined above, in such a pre-MCI SCD patient group, measures of SCD should be predictive of incident AD dementia and of biomarker abnormality indicative of AD while associations between memory performance and these outcomes should be weak or even absent. Indeed, application of the proposed neuropsychological MCI criteria of Jak and colleagues (2009) to the DCN MCI sample leads to 39% patients not meeting these more robust criteria, i.e. they would be classified as pre-MCI SCD. In these patients, short term rate of conversion to incident AD dementia, within the average 28 months follow-up interval of the DCN study, is low in (only 4%). However, a substantial number (30%) of these patients have a CSF profile indicative of AD suggesting that, given a longer follow-up, a high number of these will eventually develop robust MCI or AD dementia. In addition, as proposed above, presence of an AD biomarker signature in these patients is predicted by SCD measures while both neuropsychological measures and informant reports are not predictive (unpublished data).
An implication of these results is that new studies with longer follow-up intervals and a more extensive SCD assessment are needed to contribute further to the characterization of the pre-MCI SCD stage. In these new prospective studies it could then be examined whether more precise SCD operationalizations together with innovative objective tests (Rentz et al. 2013) can help to detect those individuals among the pre-MCI SCD patient group who have truly underlying AD pathology and will later convert to AD dementia. Such studies have only recently been set up (e.g. DELCODE or the ADNI-2 cohort), so there is only limited data today.