Abstract
Background:
Subjective memory complaints and family history of dementia are possibly intertwined risk factors for the own subsequent dementia risk and Alzheimer’s disease. However, their interaction has rarely been studied.
Objective:
To study the association between subjective memory complaints and family history of dementia with regard to the own subsequent risk of dementia.
Methods:
Cross-sectional and longitudinal analyses over a follow-up period of up to 13 years were conducted in a population sample of participants without dementia at baseline (n = 3,256, mean age = 79.62 years), using group comparisons and Cox proportional hazards models.
Results:
Cross-sectionally, participants with subjective memory complaints were significantly more likely to report family history of dementia. Longitudinally, family history of dementia (FH) was significantly associated with subsequent dementia in the subjective memory complaints (SMC) group, but not in those without SMC. A relative excess risk due to interaction analysis confirmed a significant FHxSMC-interaction.
Conclusions:
Family history of dementia was a predictor of incident dementia in those with SMC, which can serve as an additional, clinically relevant criterion to gauge the risk of dementia in older-aged subjects with SMC with and without objective cognitive impairment.
Keywords
INTRODUCTION
A first-degree relative with dementia of the Alzheimer’s type (DAT) is a risk factor to develop dementia [1, 2] beside the presence of an apolipoprotein E ɛ4 allele (APOE ɛ4) [3, 4], increasing age, and a lower educational level. Subjective cognitive decline (SCD) is a further risk factor for subsequent DAT and dementia [5]. Concerns and worries associated with SCD have been identified as particularly relevant for the prediction of subsequent dementia [6]. These worries might at least partly reflect memories of family members affected by dementia and might be associated with the assumption that the own fate will be similar. Although family history of dementia might be one reason that leads to subjective concerns, only few studies examined the association between subjective memory complaints (SMC) and family history, yet with mixed results. Haussmann et al. [7] found more subjective memory impairment and left hemispheric hippocampal cortical thinning in healthy older adults with family history of DAT. Tsai et al. [8] found that first-degree biological kinship (compared to spouses) and depressive symptoms predicted SMC in a sample of cognitively normal relatives of patients with Alzheimer’s disease. Other authors did not find associations between SMC and family history of dementia [9 –12], but associations were found between SMC and cognitive performance in early-onset DAT relatives compared to controls [13, 14]. SMC and awareness were increased in a sample of individuals with autosomal dominant DAT, but awareness decreased with the beginning of mild cognitive impairment (MCI; i.e., onset of anosognosia) in a study of Vannini et al. [15]. Heun et al. [16] found that subjective memory impairment was a better predictor for subsequent DAT than family history of DAT.
In sum, family history can be seen as an indicator of genetic risk and shared environment, but also as a psychological factor increasing personal sensitivity to age- or pathology-related changes. In the absence of objective deficits reaching the level of MCI, SMC is now frequently conceptualized as SCD in recent DAT research [5]. As we also included participants with objective cognitive impairment without dementia, we will use the term SMC in the following when referring to the results of our present study. To the best of our knowledge, family history of dementia and SMC in conjunction with genetics represented by APOE ɛ4 have not been systematically examined with regard to subsequent dementia risk. Several risk features suggested for coding in SCD studies have been proposed [17]. Family history might be a candidate for a further relevant feature in SCD studies [18]. However, memory concerns also predict risk of dementia in MCI [19]. Hence, to increase the clinical utility (i.e., regarding cases without dementia with unknown cognitive status, e.g., in family medicine practices), we examined the association between family history, SMC, and subsequent dementia risk in the present study including participants with SMC with different cognitive status except dementia. As SMC is subjective in nature, it is associated with the peculiarities of self-report measures. Additionally, SMC can be a state or trait variable and it might be associated to other factors such as depressive symptoms, feelings of worry or anxiety, and sociodemographic variables, such as, for example, education.
MATERIALS AND METHODS
Sample
We present data from a prospective multi-center study [German Study on Ageing, Cognition and Dementia in Primary Care Patients (AgeCoDe), Needs, health service use, costs and health-related quality of life in a large sample of oldest-old primary care patients (85+) (AgeQualiDe)]. At baseline, participants at least 75 years of age were recruited via general practitioners in six German cities (Bonn, Düsseldorf, Hamburg, Leipzig, Mannheim, Munich). Further inclusion criteria at baseline were absence of dementia in the general practitioner’s view and at least one contact with the general practitioner within the last 12 months. Exclusion criteria at baseline were consultations only by home visits, residence in a nursing home, severe illness the general practitioner would deem fatal within 3 months, insufficient facility in German, deafness or blindness, lacking ability to consent, and not being a regular patient of the participating practice. Written informed consent was obtained from all participants. The local ethics committees approved the study, and it was conducted in accord with the Helsinki Declaration of 1975. The interviews were conducted in-person by trained research assistants. Follow-up assessments after baseline in 2003/2004 were conducted every 18 months. Assessments after follow-up 7 were conducted every 10 months. The inclusion of participants into the study sample was completed at baseline, so that no new participants were included throughout the follow-up assessments. The present study included data of covariates and predictors assessed at baseline and information of dementia status until follow-up 9. The sample initially consisted of 3,327 participants. Participants without a dementia diagnosis at baseline were included (resulting in n = 3,256). Few participants were not included in longitudinal analyses due to the lack of information on dementia status until follow-up 9 (n = 17).
Covariates
Age (at baseline in years), sex, and education were included as covariates in the regression models. Sex was documented at baseline as female or male. Education was classified as low, medium, or high according to the revised version of the international new CASMIN educational classification [20].
Subjective memory complaints (SMC)
SMC was assessed by asking: “Do you feel like your memory is becoming worse?”. Participants could answer “no” or “yes”. If the question was affirmed, participants were asked for related worries and could state “yes, but this does not worry me” or “yes, this worries me”. SMC variables with two (no versus yes) or three levels (no; yes, without worries; yes, with worries) were derived from response to this set of questions.
Family history of dementia (FH)
Participants were asked whether their mother or their father had significant memory disorders resulting in being confused, so that they had needed someone to take care of them. They could answer “yes”, “no”, or “don’t know”. Only cases with clear answers (“yes” or “no”) for both parents were considered. A dichotomous variable of family history of dementia was built as follows: Positive answers (“yes”) of at least one parent were coded as presence of family history of dementia. Age and age of death were combined in one question that was separately assessed for mother and father (“how old is your mother/father” or “how old was your mother/father, when she/he died”). Since dementia risk correlates with increasing age, we only included those cases with a negative answer (“no”) as absent family history of dementia, if the respective maternal age or age of death was at least 75 years and the respective paternal age or age of death was at least 70 years. This procedure is in accordance with previous studies on family history [21 –23]. As the parental age restriction of the definition of a negative family history of dementia resulted in a high attrition rate of over 50%, we also included absent family history irrespective of parental age in supplementary analyses.
Genetics
Leucocyte DNA was isolated with the Qiagen blood isolation kit according to the manufacturer’s instructions (Qiagen, Germany). Further information on APOE ɛ4 genotyping can also be found in Luck et al. [24]. APOE ɛ4 status was considered as a dichotomous variable, i.e., present (at least one ɛ4 allele) or absent (without ɛ4 allele).
Dementia diagnoses
Dementia diagnoses were based on a Structured Interview for the Diagnosis of dementia of the Alzheimer type, Multi-infarct (or vascular) dementia and dementias of other etiology according to DSM-III-R, DSM-IV, and ICD-10 (SIDAM; [25]), implemented by trained research assistants. The SIDAM consists of a cognitive test battery (55 items including the Mini-Mental State Examination and covering the cognitive domains of orientation, memory, abstract reasoning, verbal ability and calculation, constructional ability, aphasia, and apraxia) and a section for clinical diagnostic impression and rating of psychosocial impairment with a scale for the assessment of activities of daily living (also cf. [24]). If the SIDAM, which included a cognitive screening, could not be assessed, a comprehensive structured proxy interview, containing a Global Deterioration Scale [26] score of at least four and/or a Blessed Dementia Rating Scale [27] score over eight, were used as a basis for dementia diagnoses. Incident dementia cases were validated by geriatric experts. All-cause dementia (i.e., DAT, vascular dementia, mixed forms, specific types, and dementia not otherwise specified) diagnosed from follow-up 1 until follow-up 9 was used as outcome variable.
Statistical analyses
In cross-sectional analyses, we examined the association between SMC, family history of dementia, and APOE ɛ4 status, in participants with available baseline data, using chi-square tests for categorical group comparisons and Kruskal-Wallis H-tests for ordinal group comparisons. Non-parametric statistics were used because a metric scale level was not assumed. In longitudinal analyses, predictor variables and covariates at baseline were used to predict the transition to all-cause dementia until follow-up 9 in Cox proportional hazards models. The models were stratified by SMC in 2 or 3 groups and adjusted for covariates in the following steps: model 1 = family history of dementia (unadjusted); model 2 = effect of family history, adjusted for age (of participants), education, and sex; model 3 = additionally adjusted for APOE ɛ4. Supplementary analyses were conducted without considering the parental age to define the absence of a family history of dementia to reduce selection bias. Level of significance was set to α< 0.05. The analyses were performed using SPSS version 25. In addition to the stratified analyses, we also computed the relative excess risk due to interaction (RERI) to examine FH × SMC interaction effects on an additive scale as implemented in the R package epiR (adjusted for age of participants, sex, and education).
RESULTS
Cross-sectional group comparisons at baseline showed a significant association between SMC in two groups (no versus yes) with family history of dementia and with APOE ɛ4 status. More specifically, participants with SMC were more likely to report a family history of dementia (χ2 (1, N = 1,240) = 7.117, p = 0.008) and were more likely to carry at least one APOE ɛ4 allele (χ2 (1, N = 3,130) = 4.460, p = 0.035), compared to participants without SMC. A significant cross-sectional association was also found between SMC in three groups and family history of dementia (χ2 (2, N = 1,240) = 19.712, p < 0.001). The association between SMC (in three groups) and APOE ɛ4 was not significant (χ2 (2, N = 3130) = 5.699, p = 0.058). Regarding family history of dementia, cross-sectional group comparisons showed a significant association with APOE ɛ4 status. More specifically, participants with family history of dementia were more likely to carry at least one APOE ɛ4 allele (χ2 (1, N = 1,209) = 14.463, p < 0.001).
In longitudinal analyses, Cox proportional hazards models were used to predict transition to all-cause dementia. Sample characteristics of longitudinal analyses are given in Table 1.
Sample characteristics of participants cross-sectionally (given in italics) and longitudinally (i.e., with or without dementia diagnosis at follow-up 9)
FUP, follow-up; M, mean; SD, standard deviation; FH, family history of dementia (unclear cases of absence of family history due to parental age were excluded); SMC, subjective memory complaints; APOE ɛ4, Apolipoprotein E ɛ4 allele; MMSE, Mini-Mental State Examination.
Stratified analyses showed that family history of dementia was associated with subsequent risk for dementia in participants with SMC (HR range, 1.54–1.74), but not in those without SMC (HR range, 1.02–1.13), in the unadjusted and also in adjusted models controlling for age, education, sex, and APOE ɛ4 (Table 2). Similar results were found when no minimum age was used to define the absence of a family history of dementia (HR for FH ranging between 0.94–1.04 in group without SMC and between 1.38–1.50 in group with SMC; also see Supplementary Table 1).
Longitudinal prediction of subsequent all-cause dementia by family history of dementia and covariates assessed at baseline stratified by SMC (in 2 groups) using a Cox proportional model
Model 1: unadjusted; model 2: age, education, sex; model 3: +APOE ɛ4; APOE ɛ4, apolipoprotein E ɛ4; SMC, subjective memory complaints; HR, hazard ratio; CI, confidence interval; age, age at baseline (in years).
In addition to the presented stratified analyses, a RERI analysis also provided evidence confirming a significant FH × SMC interaction effect on an additive scale (estimate = 0.8361879, lower CI = 0.1571014, upper CI = 1.515274).
Additionally, SMC in three groups (no; yes, without worries; yes, with worries) was analyzed (Table 3). In these stratified analyses, family history of dementia was only associated with the participants' subsequent dementia risk in the group of participants with SMC with worries. The association in SMC with worries remained significant after controlling for age, education, and sex (Table 3, model 2; HR = 2.06, 95% CI = 1.30–3.27) and also after additional control for APOE ɛ4 (Table 3, model 3; HR = 1.89, 95% CI = 1.19–3.01). Only a trend for an association between family history of dementia and own subsequent dementia risk was found for SMC without worries (HR for FH ranging between 1.32–1.48 in models 1 to 3, see Table 3).
Longitudinal prediction of subsequent all-cause dementia by family history of dementia and covariates assessed at baseline stratified by SMC (in 3 groups) using a Cox proportional model
Model 1: unadjusted; model 2: age, education, sex; model 3: +APOE ɛ4; APOE ɛ4, apolipoprotein E ɛ4; SMC, subjective memory complaints; HR, hazard ratio; CI, confidence interval; age, age at baseline (in years).
In the supplementary analysis where no minimum age was used to define the absence of a family history of dementia, the association between family history of dementia and subsequent dementia was also not significant in the group without SMC and partly significant in SMC groups, but not more pronounced in SMC with worries (adjusted for age, education, and sex; family history of dementia in group without SMC: HR = 1.04, 95% CI = 0.69–1.57; family history of dementia in group with SMC without worries: HR = 1.41, 95% CI = 1.04–1.92; family history of dementia in group with SMC with worries: HR = 1.46, 95% CI = 1.04–2.05; adjusted for age, education, sex, and APOE ɛ4; family history of dementia in group without SMC: HR = 1.04, 95% CI = 0.69–1.57; family history of dementia in group with SMC without worries: HR = 1.35, 95% CI = 0.99–1.83; family history of dementia in group with SMC with worries: HR = 1.31, 95% CI = 0.93–1.83; also see Supplementary Table 2).
DISCUSSION
Family history of dementia, genetic variants related to dementia, and SMC all increase the risk for dementia and are probably related to each other. Psychologically, the experience of close relatives being affected with dementia may give rise to closer self-monitoring of cognitive function in older adults. Family history is also a proxy for genetic risk factors, including APOE ɛ4 and genetic susceptibility variants identified in genome-wide association studies, all modulating cognitive performance along different clinical stages.
Cross-sectional group comparisons at baseline showed a significant association between SMC and family history of dementia. Participants with SMC were more likely to report a family history of dementia. The group of participants with a family history of dementia and the group of participants with SMC were both more likely to carry at least one APOE ɛ4 allele. Furthermore, the size of the hazard ratios that we found in the whole group for the association between family history of dementia and subsequent own dementia risk (unadjusted HR = 1.38, HR adjusted for age, education, and sex = 1.57) of about HR = 1.6 were comparable to another population-based cohort study without stratification for SMC [2]. Considering those with SMC in more detail, we observed that their dementia risk was nearly doubled by a family history of dementia in those who had SMC with worries, while in those without worries, the risk increase was smaller, about 50%. Together, these findings suggest that family history of dementia contributes to the risk of progression to dementia in participants of about 80 years on the whole cognitive continuum except dementia with SMC, and in part this will be due to biological factors. Importantly, compelling evidence supports the biology behind the easily assessable, non-invasive variable of family history (also cf. [28, 29]).
Beyond biological and genetic validity, family history can also capture environmental risk or protective factors such as socioeconomic status, diet, or social interaction styles (cf. [4, 30]). Cognitive concerns might in part be psychologically-driven based on experiences, memories or stories within the family. While SMC initially might be “biologically” triggered by neurodegenerative processes ongoing in these individuals, a higher awareness of such deficits might result from the experienced family history of dementia and might in turn aggravate concerns, potentially triggering a vicious and self-reinforcing circle of worries, awareness, and self-attention. Supporting this idea, Ostergren et al. [31] reported that family history of DAT and poorer self-rated memory was associated with higher perceived threat of DAT. Moreover, repetitive negative thinking as a more general thinking pattern of worry (relating to future) and rumination (relating to past), possibly also related to cognitive complaints, was associated with cognitive decline and DAT biomarkers in a cognitively healthy sample with a family history of DAT [32].
However, the fact that family history of dementia was still associated with subsequent dementia risk in subjects with SMC after controlling for APOE ɛ4, also suggests that a family history of dementia contains information beyond genetics. Shared lifestyle factors, like nutrition, socioeconomic status, living in the same geographic region with certain environmental risk factors, heritable personality traits, and social interaction styles are shared in families, and all of these could contribute to the effect, if they are related to dementia risk. As expected, APOE ɛ4 and (higher) age of participants were associated with an increased risk of subsequent dementia, whereas education and sex showed no consistent significant association with the subsequent dementia risk.
Strengths and limitations
Our study holds several strengths. We present novel data on the joint association of family history of dementia, SMC, and genetics with subsequent risk for dementia in a population-based cohort study. We considered participants with different cognitive status except initial dementia which increases the external validity of our results and its clinical usability.
Family history was assessed by self-reports of the participants and could not be validated by medical registry-based data or neuropathological autopsy-based examinations. Parental dementia is usually a stressful experience not to be forgotten. Nonetheless, the fact that we assessed family history of dementia not by informants or medical records, but directly asked our study participants themselves might have affected the validity of our data, although we excluded participants with a prevalent dementia diagnosis at baseline. However, several previous studies provided evidence for an acceptable accuracy of self-reported data on family history of dementia [33, 34]. In line with these findings, the information obtained this way was a valid indicator of risk in our study.
Besides this, we combined parental family history of dementia into one single variable and considered this reasonable due to sample size issues. Differences between maternal and paternal family dementia history have been reported, with a stronger maternal transmission reported [1, 35], although others did not find specific parental effects [2].
Our study is not without limitations. Due to the fact that the risk for dementia rises with increasing age, the age of death of parents needs to be considered. Consequently, we defined a negative family history of dementia with a maternal age or an age of death of at least 75 years, and a paternal age or an age of death of at least 70 years. Unfortunately, the parental age restriction of the definition of a negative family history of dementia in combination with other missing variables resulted in a very high rate of about 60% of data missing regarding family history. A selection bias, especially with regard to the group without a family history of dementia, might endanger the generalizability of our results, as participants whose mothers reached an age of at least 75 years and fathers who reached an age of at least 70 years particularly in our cohort might not be representative. The type of family history dementia is relevant in predicting future dementia risk but was unknown in the present study. Additionally, our results may not be valid for younger age cohorts.
We used all-cause dementia in the family history of dementia and in the incident subsequent dementia diagnosis variable, although APOE ɛ4 is a rather DAT-related genetic risk factor. However, as Alzheimer’s disease should be the most frequent etiology in the age segment of our sample, we do not consider this limitation as too severe. In fact, we observed 752 incident cases of all-cause dementia until follow-up 9, 515 of which were clinically consistent with DAT (also including mixed forms) which approximates about 70% of our cases. Further epidemiological information on the study sample, such as conversion rates, can be found in the literature [36, 37].
Conclusions
Considering our results, we propose that a family history of dementia might be added to the list of features suggested for coding in SCD studies, analogous to APOE genotype, as we illustrated, after controlling for other factors, FH increased the risk of dementia in persons with SMC. In addition, SMC might qualify as important additional information to further specify the risk for dementia emanating from family history in participants of advanced age. The latter is based on the observation that participants without SMC showed no significant association between family history of dementia and own subsequent dementia risk. In other words, without SMC, a positive family history might not be a relevant predictor for future dementia in subjects over 75. Finally, we propose that participants with SMC, particularly those with worries, in addition to affected first-degree relatives with dementia can be offered counselling or further diagnostics such as neuropsychological assessments, depending on the individual risk.
Footnotes
ACKNOWLEDGMENTS
We want to thank both all participating patients and their general practitioners for their good collaboration.
Principal Investigators*: Wolfgang Maier, Martin Scherer, Steffi G. Riedel-Heller
Heinz-Harald Abholz, Christian Brettschneider, Cadja Bachmann, Horst Bickel, Wolfgang Blank, Hendrik van den Bussche, Sandra Eifflaender-Gorfer, Marion Eisele, Annette Ernst, Angela Fuchs, André Hajek, Kathrin Heser, Frank Jessen, Hanna Kaduszkiewicz, Teresa Kaufeler, Mirjam Köhler, Hans-Helmut König, Alexander Koppara, Diana Lubisch, Tobias Luck, Dagmar Lühmann, Melanie Luppa, Tina Mallon, Manfred Mayer, Edelgard Mösch, Michael Pentzek, Jana Prokein, Alfredo Ramirez, Susanne Röhr, Anna Schumacher, Janine Stein, Susanne Steinmann, Franziska Tebarth, Carolin van der Leeden, Michael Wagner, Klaus Weckbecker, Dagmar Weeg, Jochen Werle, Siegfried Weyerer, Birgitt Wiese, Steffen Wolfsgruber, Thomas Zimmermann.
*Hendrik van den Bussche (2002–2011).
FUNDING
This study/publication is part of the German Research Network on Dementia (KND), the German Research Network on Degenerative Dementia (KNDD; German Study on Ageing, Cognition and Dementia in Primary Care Patients; AgeCoDe), and the Health Service Research Initiative (Study on Needs, health service use, costs and health-related quality of life in a large sample of oldest-old primary care patients (85+; AgeQualiDe)) and was funded by the German Federal Ministry of Education and Research [grants KND: 01GI0102, 01GI0420, 01GI0422, 01GI0423, 01GI0429, 01GI0431, 01GI0433, 01GI0434; grants KNDD: 01GI0710, 01GI0711, 01GI0712, 01GI0713, 01GI0714, 01GI0715, 01GI0716; grants Health Service Research Initiative: 01GY1322A, 01GY1322B, 01GY1322C, 01GY1322D, 01GY1322E, 01GY1322F, 01GY1322G]. Genotyping of AgeCoDe was funded by the EU Joint Programme –Neurodegenerative Disease Research (JPND) EADB grant [BMBF grant: 01ED1619A] to Alfredo Ramirez. Alfredo Ramirez is also supported by the German Research Foundation (DFG) [grants Nr: RA 1971/6-1, RA1971/7-1, and RA 1971/8-1].
The funding sources were not involved in study design; in the collection, analysis and interpretation of data; in the writing of the report; and in the decision to submit the article for publication.
CONFLICT OF INTEREST
The authors have no conflict of interest to report.
DATA AVAILABILITY
Data is not publicly available, but reasonable requests may be sent to the corresponding author to be transferred to the Principal Investigators of the study group.
