Abstract
Aim:
To assess memory impairment insight as a predictor of dementia and Alzheimer’s disease (AD) in amnestic mild cognitive impairment (MCI).
Methods:
To verify whether the awareness of memory impairment assessed by Geriatric Depression Scale (GDS) was associated with the risk of progression to dementia and AD in a cohort of MCI, we used a Cox regression model adjusted for age, sex, education, subtypes of amnestic MCI, Mini-Mental State Examination, Cumulative Illness Rating Scale severity index, and apolipoprotein E genotype.
Results:
During a follow-up of 27.7 (20.8) months, 205 (63.3%) of 324 patients with amnestic MCI progressed to dementia, including 141 to AD. No association was found in the unadjusted, partially adjusted (for sociodemographic variables), and fully adjusted multivariate Cox analysis between the awareness of memory impairment and the progression to dementia and AD.
Discussion:
Awareness or anosognosia of memory deficits, identified by GDS, is not useful to predict progression to dementia of patients with amnestic MCI.
Introduction
Self-awareness may be defined as “the capacity to perceive the ‘self’ in relatively objective terms while maintaining a sense of subjectivity.” 1 In the clinical neurological literature, the term “anosognosia” is used to describe the difficulty to appreciate or evaluate a sensory motor, behavioral or cognitive deficit, and the related functional effects. 2 In the context of dementia, it is intended as a lack of insight in impairments affecting activities of daily living or in cognitive dysfunction, particularly the memory domain. 3
Low awareness of memory deficit in patients affected by dementia has been well documented 4 and has been mainly studied in the context of Alzheimer’s disease (AD) in which the prevalence dramatically varies depending on the measurement tools and setting, ranging from 30% to 80%. 5
There is less information about the frequency and correlates of anosognosia in patients with mild cognitive impairment (MCI), 6 a heterogeneous clinical and research entity described as an intermediate stage between normal cognition and dementia, with an objective abnormality of the person’s cognitive capacities not interfering with his/her functional independence. The MCI is considered a relevant risk condition for incident dementia. In Peterson’s early MCI criteria, 7 the presence of subjective memory complaint was considered necessary, but the need of a concurrent confirmation by caregivers was later highlighted by Petersen himself, so that more recently subjective perceived cognitive issues were not necessarily required anymore to MCI diagnosis, 8 due to reduced reliability of patient’s self-report. The predictive validity of self-reported memory complaints for memory impairment and dementia is still a matter of debate. The self-reported cognitive complaint has usually strong implications during the very early stages of the cognitive decline, representing also a predictor for the subsequent development of cognitive impairment and dementia. Indeed, a meta-analysis of 2014 reported an overall 24.4% rate of progression to MCI and a 10.9% to dementia in patients with subjective memory complaints considering studies over 4 years. 9
Obviously, it is arduous to study a complex phenomenon such as awareness. First of all, it is a multidimensional concept which can be approached with several assessment tools. Difficulty in insight could concern different cognitive domains, functional, behavioral, or affective status, and reflect discrepancy between a personal belief compared with an informant report or an objective measure. In 2009, Roberts et al 10 analyzed in a systematic review all available studies which had evaluated the awareness level in patients with MCI in either a clinical or research setting. The authors concluded presenting evidence for great variability in the level of awareness of memory deficit among individuals with MCI. A more recent review of self-awareness in MCI, 11 quantitatively summarizing the evidence of impaired awareness in MCI, added that the level of awareness may vary according to cognitive impairment and memory and language capacities. Few data are available on predictivity of awareness of impairment in MCI. Tabert et al 12 found that unawareness of functional impairment in patients with MCI predicted progression to dementia. A prospective study 13 has focused on anosognosia of cognitive deficit in MCI and showed that low awareness of memory deficit in amnestic MCI predicts conversion to dementia; to our knowledge, no other studies investigated insight of cognitive deficits in MCI as a predictor of progression to dementia.
In order to test whether an association between unawareness of memory deficit at baseline and progression of MCI to dementia exists, we followed up a cohort of patients with MCI in our memory clinic.
Materials and Methods
Participants
The participants of the present study were recruited among a cohort of patients with MCI attending our outpatient memory clinic in a University Hospital of Milan. Informed written consent was obtained from all participants. The local ethics committee approved the study protocol. Between January 2003 and May 2014, 469 participants attending to our service met the diagnostic criteria for MCI.
Baseline Assessment
At baseline, all study participants underwent an extensive evaluation following a standardized diagnostic protocol. This included medical and neurological examinations, laboratory testing, neuroimaging (computed tomography scan or magnetic resonance imaging), and neuropsychological assessment, which included Mini-Mental State Examination (MMSE) 14 as test of general cognition and an extensive neuropsychological battery, as previously described. 15 The apolipoprotein E (APOE) genotype was available in a subgroup of 234 (72%) patients.
Somatic comorbidities were quantified by the modified Cumulative Illness Rating Scale (CIRS). 16 The CIRS severity index, which is the mean of the scores of the first 13 categories (excluding psychiatric diseases), was also calculated and used to estimate the impact of comorbidities.
The presence of depression was mainly assessed using the 30-item version of the Geriatric Depression Scale (GDS) and defined with a cutoff score for depression ≥11. 17 To assess the awareness of memory impairment, we retrospectively analyzed the answer, namely, yes or no, provided at baseline evaluation to a specific question included in GDS: “Do you feel you have more problems with memory than most people?”
Definition and Subtypes of MCI
Participants were classified as MCI 8 if they had subjective cognitive complaints and/or informant report, objective cognitive impairment in at least 1 task of an extensive neuropsychological assessment, 15 essentially preserved daily functioning (defined as full score in basic activities of daily living 18 and full score or minimally impaired instrumental activities of daily living 19 ), and absence of dementia defined according to Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria. Exclusion criteria were the presence of psychiatric disorders that clearly preceded the onset of cognitive impairment, the presence of major neurologic illness (eg, Parkinson disease, amyotrophic lateral sclerosis, stroke), a history of significant traumatic brain injuries, and current or past alcohol or substance abuse.
The MCI cases were classified into the following subtypes: 6 (1) amnestic MCI, if there was an isolated memory impairment; (2) single-domain nonamnestic MCI, if there was an isolated impairment of a cognitive domain other than memory; (3) multiple-domain amnestic MCI characterized by an impairment of multiple cognitive domains, including memory; and (4) multiple-domain nonamnestic MCI, with an impairment of multiple cognitive domains but without memory deficits. Impairment at neuropsychological tests was defined by a score = 0 at the 5-point scale of equivalent scores, which is the method of standardization of almost all neuropsychological tests in Italy. 20,21 In the present study, we enrolled only amnestic MCI (single and multiple domain) cases.
Follow-Up
Patients were reassessed at 12-month intervals to monitor the course of MCI, by neuropsychological tests and clinical examination. The length of follow-up was expressed in person-years and calculated as the time interval from baseline assessment to the last follow-up for the patients who did not develop dementia and as the time interval from baseline assessment to the time of diagnosis for the patients who progressed to dementia. The surveillance ended at the diagnosis of dementia or death. May 2015 was set as the end of the cohort surveillance.
Outcome
Standard criteria 22 defined diagnosis of dementia. Patients diagnosed with dementia were classified according to standardized criteria, based on the type of dementia, as having AD, 23 Lewy body dementia (LBD), 24 frontotemporal dementia, 25 and vascular dementia (VaD). 26 For patients who refused the follow-up visit, ascertainment of dementia was performed through an informant telephone interview, following the validated informant-based Clinical Dementia Rating Scale score. 27
Statistical Analysis
Continuous measures were reported as mean and standard deviation and were analyzed with Student t test. Categorical variables were reported as absolute number and percentage and were compared with the Pearson χ2 test. Patients were considered at risk of dementia until one of the following outcomes occurred: (1) incident dementia, (2) death, and (3) end of the study.
Independent Cox proportional hazards regression models were used to estimate the relative risk of progressing to dementia in relation to the awareness of memory impairment assessed by GDS. Different models were carried out: (1) an unadjusted model; (2) a partially adjusted model for sociodemographic variables (age, gender, and education as categorical variables); (3) a multiadjusted model for sociodemographic variables, MMSE score (as categorical variable: <24 vs ≥24), 28 GDS score (as a continuous variable), MCI subtype, and CIRS severity index; and (4) a fully multiadjusted model for all abovementioned variables of the multiadjusted model and also for APOE genotype. These variables were included as they all could have confounded or modified the longitudinal relationship between awareness of memory impairment and dementia. The same models and analytic strategy were repeated using AD rather than dementia as an outcome. All analyses were performed using SPSS 21 with an α level of P < .05.
Results
Between January 2003 and May 2014, we included 348 amnestic MCI (single or multiple domain) attending our memory clinic (Figure 1). We lost 24 patients at follow-up (6.9%); dropout participants did not differ in any characteristics from the study participants. The cohort consisted of 324 patients (55.2% female), with a mean age of 75.3 (6.8) years and a mean MMSE score of 24.9 (2.8).
Flow chart of study participants.
During the 746.5 person-years of follow-up (minimum 0.3, maximum 10.8), 119 (36.7%) patients did not progress to dementia. There were 205 (63.3%) incident dementia cases, including 141 (68.8%) AD, 28 (13.6%) mixed degenerative–vascular–carential dementia, 16 (7.8%) VaD, and 6 (3.0%) LBD. Fourteen (6.8%) patients received an unspecified dementia diagnosis by phone interview (Clinical Dementia Rating Scale ≥1). The GDS score was available for 282 patients, which represented the final cohort, whose baseline characteristics are shown in Table 1. We did not find any demographic or clinical statistically significant difference between people who received GDS (87%) and patients who were investigated with other depression rating scales.
Demographic and clinical characteristics at baseline of whole sample and by awareness of their memory impairment.a
Abbreviations: AD, Alzheimer’s disease; APOE, apolipoprotein E; CIRS, Cumulative Illness Rating Scale; GDS, Geriatric Depression Scale; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; NS, not significant; SD, standard deviation.
aFigures in parentheses are percentages.
bCalculated on a subgroup of 209 patients with MCI.
Depressed MCI (GDS ≥ 11) were more likely to be aware of their memory deficits (54.2% vs 28%; P < .001). Age, sex, level of education, MMSE score, MCI subtype, CIRS severity index, APOE genotype, and progression to dementia and AD did not differ between patients aware and unaware of their memory difficulties. Stratifying our cohort by age into 2 groups (separated by the median value = 76.2 years) or into 3 groups with equal number of patients and comparing the first to the last group, we did not find any difference in awareness of memory difficulties.
Tables 2 and 3 report hazard ratio and 95% confidence intervals of unadjusted, partially adjusted, multiadjusted, and fully multiadjusted Cox proportional hazards regression models to estimate the relative risk of progressing, respectively, to dementia and AD according to awareness of memory impairment in patients with amnestic MCI. Results showed that being older and homozygous for ∊4 allele significantly increase the risk of dementia and AD. An MMSE score ≥24 and a higher CIRS severity index reduced the risk of dementia and AD. In all the models, neither awareness nor anosognosia of memory impairment significantly modified the risk of progression to dementia and AD.
Risk of developing dementia according to awareness of memory impairment in MCI subjects.a
Abbreviations: APOE, apolipoprotein E; CIRS, Cumulative Illness Rating Scale; GDS, Geriatric Depression Scale; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination.
aAll values are expressed as hazard ratio (95% confidence interval).
b P < .05.
c P < .001.
Risk of developing Alzheimer’s disease according to awareness of memory impairment in MCI subjects.a
Abbreviations: APOE, apolipoprotein E; CIRS, Cumulative Illness Rating Scale; GDS, Geriatric Depression Scale; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination.
aAll values are expressed as hazard ratio (95% confidence interval).
b P < .05.
c P < .001.
Discussion
In this clinically based study, we did not find any statistically significant association between awareness of memory impairment (as assessed with an item of GDS) and the progression to dementia in a cohort of participants with MCI.
To our knowledge, only one previous clinical longitudinal study had prospectively evaluated the awareness of memory impairment in amnestic MCI. 13 In particular, Spalletta and colleagues, aiming mainly to describe neuroanatomical correlates of this phenomenon, found higher difficulties in appraising memory deficits in patients who would convert to dementia compared to nonconverters. This result was partly limited by the small experimental sample, including 36 amnestic MCI.
Several other studies have investigated the occurrence of subjective memory complaints in patients with MCI compared to healthy ones 10 . Overall, the results appear scarcely comparable, due to the differences in the setting (population vs clinical study), the diagnostic criteria employed and the way they were operationalized (inconsistent neuropsychological measures and cutoffs), and the evaluation of awareness. With these premises, the level of awareness in MCI seems to vary, with some suggestion that a proportion of people with MCI may tend to overestimate their dysfunction. This last finding could result from disparity between reports from patients with MCI and from informants or could be due to factors other than awareness.
Several mechanisms could explain our findings. Probably, the lack of insight into memory impairment might develop in later stages of the cognitive degeneration and could explain the relationship between dementia severity and anosognosia described in other studies. 29 This hypothesis is consistent also with the results of two other studies, 3,30 which found that anosognosia, measured by a discrepancy between self and caregiver viewpoint, is a frequent symptom in very mild AD, but not in patients with MCI. In particular, in the study of Kalbe et al, 30 it has been suggested that MCI could have a decrease in this insight with the progression of the disease. Furthermore, the presence of subjective memory complaints could be at least partly due to the presence of depression and our results are in line with this hypothesis: we have found a positive relation between awareness of memory deficit and a score suggestive of depression at GDS. Previous studies conducted on AD had shown that patients who were more depressed demonstrated increased insight of disease. 31 -33 Similar results were found also in a healthy elderly patients: Minett et al 34 found a stronger association between subjective memory complaints and depressive symptoms rather than poor cognitive performance. Schmand et al 35 suggested that the relationship between memory complaints and depression could be reciprocal: complaints about memory could derive from depression which may enhance negative attributions such as emphasizing memory issues; conversely, recognition of failing memory could lead to depression.
Anosognosia of memory functioning could coexist with apathy. Mak et al 36 reported an association between anosognosia and apathy in AD, but not in patients with MCI. However, this study is limited by the small sample size (21 MCI, 36 AD).
The great variability of awareness in MCI could be due to impairment, caused by degenerative and/or vascular processes, affecting different cortical and subcortical regions of the brain involved in determining insight and also clinical depression or apathy. Interestingly, recent findings suggest that apathy and depression in patients with AD involve distinct functional circuits of frontal and temporal areas. 37 Many studies have linked anosognosia in AD with hypometabolism or hypoperfusion in orbital, lateral, and medial frontal cortex; neural correlates of defective awareness of cognitive deficits in MCI seem to involve frontal, prefrontal, and also temporal areas. 13 In particular, in the study of Spalletta et al, 13 reduced awareness of cognitive deficits in MCI who converted to dementia correlated positively with reduced gray matter volume in different areas compared with MCI stable at 5-year follow-up (cerebellar vermis in converters vs left superior temporal area in nonconverters to dementia).
Finally, self-awareness of memory impairment represents only a part of a broader concept named metamemory, 38 a framework involved in memory function, which encompasses beliefs, sensations, attitudes, and knowledge about memory function.
The diversity of definitions and the variety of research methods used to measure meta-memory dramatically affect the insight phenomenon, giving rise to incomparability across studies. With these premises, patients with amnestic single-domain MCI with anosognosia for their memory impairment should be considered as having multiple-domain MCI.
The strengths of our study are (1) an extended group of amnestic MCI, clinically and psychometrically well defined; (2) the use of a hard end point instead of surrogates (clinical diagnosis of dementia); and (3) the low dropout rate.
The findings reported in this study should be considered in the context of its limitations: (1) the heterogeneity of the construct of MCI; (2) the use of a simple question instead of a more complex questionnaire or a combination of questions of GDS, which allowed us to investigate exclusively the possible presence of insight of memory impairment—not its severity—merely in patients with amnestic MCI, without broadening the concept of our analysis; (3) the impossibility to disentangle the self-reported memory impairment with the caregiver perception of worsening in the memory performance of the patient; however, the choice of Wimblad’s 8 rather than Petersen’s criteria 6 allowed clinical follow-up and subsequently inclusion in the study of 164 patients (58% of the cohort population) otherwise disregarded; and (4) the lack of neuroimaging correlates of awareness of memory deficit.
Future studies should aim to assess awareness of cognitive function through the use of questionnaires covering various cognitive domains in order to investigate whether an extensive neuropsychological battery (maybe including a measure of discrepancy score between self- and informant rating on parallel forms of questionnaires) could provide any predictive utility in the follow-up diagnosis of MCI.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was partly supported by the “Associazione per la Ricerca sulle Demenze ONLUS.”