Risk Factors of Rapid Cognitive Decline in Alzheimer’s Disease and Mild Cognitive Impairment: A Systematic Review and Meta-Analysis

Abstract

Background:

The conclusions about risk factors for rapid cognitive decline (RCD) in Alzheimer’s disease (AD) and mild cognitive impairment (MCI) remain contradictory.

Objective:

To explore the factors predicting RCD in AD and MCI.

Methods:

We searched the PubMed, EMBASE, and the Cochrane Library from inception to May 27, 2017 for studies investigating factors associated with faster cognitive progression in AD and MCI. Effect sizes were meta-analyzed using fixed-effects and random-effects models.

Results:

Fifty-three studies with 14,330 patients (12,396 AD and 1,934 MCI) were included in the systematic review. The following factors were identified to increase the risk of RCD in AD: Apolipoprotein E4 (ApoE4) (SMD [95% CI]: 0.52 [0.06,0.98]), early age at onset (SMD [95% CI]: –0.42 [–0.71, –0.13]), high level of education (RR = 2.05, 95% CI = 1.26 to 3.33), early appearance of extrapyramidal signs (RR = 2.18; 95% CI = 1.30 to 3.67), and neuropsychiatric conditions including hallucination (RR = 2.01, 95% CI = 1.40 to 2.87), strolling (RR = 1.99, 95% CI = 1.38 to 2.86), agitation (RR = 1.66, 95% CI = 1.23 to 2.24), and psychosis (RR = 1.42, 95% CI = 1.07 to 1.89). Instead, advanced age (≥75 years) (RR = 0.96, 95% CI = 0.93 to 0.99), diabetes (RR = 0.57; 95% CI = 0.35 to 0.93), and multidrug therapy (RR = 0.61, 95% CI = 0.60 to 0.62) would lower the risk of RCD. Furthermore, systematic research also reviewed seven risk factors associated with RCD in MCI.

Conclusion:

ApoE4, early onset, early appearance of extrapyramidal signs, high education level, and neuropsychiatric conditions might increase the risk of RCD while older age, diabetes, and multidrug therapy were the protective factors for AD.

Keywords

Alzheimer’s disease meta-analysis mild cognitive impairment rapid cognitive decline risk factor systematic review

INTRODUCTION

Alzheimer’s disease (AD) is a progressive neurodegenerative disorder featuring unremitting decline of cognitive function with a devastating course in late-life [1]. AD places a heavy burden on society and families. According to the World Alzheimer Report 2016, the total expense of care service for people age ≥65 years with dementia added up to $236 billion [2]. Mild cognitive impairment (MCI) was considered as a transitional stage from normal aging to clinical dementia [3]. The rates of cognitive decline in AD and MCI showed diversity among individuals and the concept of rapid cognitive decline (RCD) was proposed to characterize the specific population exhibiting faster cognitive decline than others [1, 4]. As for the definition, most studies chose the decline of ≥3 points per year on the Mini-Mental State Examination (MMSE) as the criterion, according to which about 33.9% of AD patients were fast decliners [1, 5]. More importantly, it was shown that patients with RCD had a worse prognosis for mortality and autonomy loss [6 –8]. Therefore, it is an imperative problem for clinicians to explore risk factors predicting RCD and make ideal management measures for these patients [9].

In the past decades, some potential factors have been reported to predict the progression of cognitive decline in AD and MCI, such as sociodemographic factors, genetic factors, lifestyle, drug therapy, comorbidities, and biochemical markers in cerebrospinal fluid (CSF). Notably, previous systematic reviews have evaluated many risk factors for RCD due to AD [1 , 9–11]. However, relevant meta-analyses have never been conducted. So what are the risk factors associated with rapid cognitive decline in AD and MCI patients according to the meta-analysis?

METHODS

Search strategy

A systematic literature search was conducted in PubMed, EMBASE, and the Cochrane Library from inception to May 27, 2017 by two investigators to identify relevant studies, which followed the current standards in PRISMA 2009 guidelines (Supplementary Table 1). The search terms include Alzheimer’s disease, dementia, mild cognitive impairment, risk factor, rapid cognitive decline, rapid decline, fast progression, fast decliner, and survival. The detailed search strategy was listed in Supplementary Table 2. We restricted our search to those written in English. In addition, reference lists of reviews, systematic reviews, and meta-analyses were manually retrieved for additional appropriate studies.

Study selection

Studies were included if they met the following criteria: 1) the study used a prospective cohort design; 2) participants fulfill the diagnostic criteria of AD or MCI at baseline; 3) the outcomes were divided into categorical (incident RCD) and continuous (declined scores of relevant scales) variables; 4) the risk estimates were provided, including relative risks (RRs), odd ratios (ORs), hazard ratios (HRs) with 95% confidence interval (CI), mean and its standard deviation (SD), or raw data to calculate these numbers. Besides, if two publications reported the same risk factors based on the same population, we chose the latest research. We excluded case-reports, commentaries, meeting abstracts, reviews, and meta-analyses.

Data extraction and quality assessment

As for each study, the following information was extracted: the first author, publication year, country, sample size, follow-up period, mean age, percentage of females, mean baseline MMSE score, MMSE re-check frequency and follow-up times, definition of RCD, and confounders adjusted. Meanwhile, for categorical variables, estimates from both univariate and multivariate analyses were extracted to assess the influence of different analysis models on the final results.

Quality assessment of included studies was conducted using the Newcastle Ottawa Scale (NOS) whose total score is 9 points [12]. The scale is used to evaluate the quality of observational studies across three aspects, including selection, comparability, and outcome [13]. Both data extraction and quality assessment were conducted by two evaluators independently and if discrepancy appears, the conclusion would be drawn by discussing with the third individual.

Statistical analyses

Given the low incidence of RCD, ORs were directly regarded as approximate estimates of RRs. For binary data, log transformation was performed before estimating pooled RR. For continuous data, standard errors were converted to standard deviations for further analysis and the pooled effect sizes were presented in standard mean difference (SMD) and its 95% CI. The risk estimates (including RRs, mean and its standard deviation) were pooled using fix-effects and random-effects models. Heterogeneity between studies was assessed using Q test and I². The random-effects model was used when statistically significant heterogeneity was found (I² >50%, p < 0.05) [14, 15]. Publication bias was examined by Egger test and when statistically significant bias was found, the trim and fill method was employed to adjust it [16]. Sensitivity analyses to exclude those rated with lower quality scores were also conducted. All analyses above were conducted using the metan package in R software (version 3.4.1).

RESULTS

Literature searching and characteristics of included studies

Figure 1 shows a flow chart for literature search and study selection. Initially, a total of 14,397 literatures were identified. After de-duplication and retrieval of additional records, 9,772 articles were included. Eventually, 569 articles were eligible for inclusion via reviewing the titles and abstracts. After scanning the full-texts, we included 53 studies, of which 35 about AD were eligible for meta-analysis. The detailed characteristics of included studies were provided in Table 1.

Fig. 1

Flow chart of study selection process.

Table 1

The characteristics of included studies

Study	Country	Sample size	Follow-up period (SD)	Mean age years (SD)	Sex (% women)	Mean baseline MMSE score (SD)	MMSE re-check frequency	Definition of RCD	Confounders
Carcaillon et al. [1]	France	204	377.6 (61.2) days	78.0 (7.5)	119 (58.3%)	19.7 (4.4)	Once per month/13 times	≥3 points/ year (MMSE)	Residential status, level of education and loss of autonomy, loss ≥3 points on MMSE prior inclusion.
Musicco et al. [2]	Italy	154	2 years	73.4 (8.2)	101 (65.6%)	18.0 (4.4)	Once per 6 months/4 times	≥5 points/2 years (MMSE)	Age, gender, education, disease duration, vascular comorbidity, use of neuropsychiatric drugs.
Musicco et al. [3]	Rome	154	23 (15.6) months	73 (8.2)	101 (65.6%)	18 (4.8)	Once per 6 months/4 times	≥2.5 points/year (MMSE)	Age, education, severity and duration of disease, family history of dementia, comorbidity.
Canevelli et al. [4]	Europe	1005	2 years	76.1 (7.7)	64.50%	20.6 (3.9)	Once per 6 months/ 4 times	≥6 points/ year (MMSE)	NA
Sona et al. [5]	Australia	156	18 months	76.9 (8.1)	96 (61.5%)	15.4 (7.0)	Once per 18 months/Once	≥4 points/ year (MMSE)	Family history of dementia, smoking, ApoE4, comorbidities, education, ChEIs, gender.
Degerman Gunnarsson et al. 2014 [6]	Sweden	196	2–9 years	70 (46–86)	111 (57%)	24 (4)	Once per 6 months to 7 years (median = 30 months)	≥4 points/ year (MMSE)	Age, education, coronary heart disease/heart failure and baseline AD.
Lopez et al. [7]	France	95	3.6 (1.8) years	72.7 (7.2)	68 (71.6%)	17.9 (4.8)	Annually/3 times	NA	Age, education, gender, initial severity of the dementia.
Huang et al. [8]	Taiwan	80	1 year	73.9 (3.6)	59 (73.8%)	19.57 (4.6)	Once per 6 months/2 times	≥median value (CASI and EFT scores)	Cerebrovascular risk factors and nutritional factors.
Masse et al. [9]	France	342	34.8 (18.9) months	73.5 (7.3)	232 (67.8%)	21.3 (5.1)	Once per 6 to 12 months/3–6 times	≥1.8 points/ year (MMSE)	Sex, age, education, diabetes, baseline MMSE scores, continuous treatment with ChEI, or antidepressant treatment.
Lopez et al. [10]	Pennsyl- vania	270	1 year	72.6 (8.7)	33.70%	18.8 (7.3)	Annually/Once	≥2 points/ year (MMSE)	Age, sex, education, MMSE score, BDRS, ADLs, extrapyramidal signs, psychosis, depression, medication, nursing home admission.
Gillette- Guyonnet et al. [11]	France	455	1 year	75.4 (6.7)	NA	18.9 (5.4)	Once per 6 months/2 times	≥3 points/year (MMSE)	Education, initial BMI, and initial number of disabilities on the IADL.
Barbe et al. [12]	France	61	1 year	81.7 (6.6)	38 (62.3%)	21.0 (4.7)	Annually/Once	≥3 points/year (MMSE)	NA
Chui et al. [13]	California	135	26.5 weeks	72.9 (8.1)	105 (77.8%)	19.6 (3.7)	Once per month/6 times	≥6 points/year (MMSE)	Initial severity of dementia.
Stern et al. [14]	USA	236	6.9 (9.2) years	73.1 (8.9)	140 (59.3%)	mMMS 37.9 (5.6)	Once per 6 months/12 times	NA	Age, sex, disease severity, and estimated duration of illness.
Scarmeas et al. [15]	USA and Europe	497	4.4 (3.1) years	73.8 (8.9)	300 (60%)	20.4 (3.7)	Once per 6 months/8.8 times	NA	Cohort, recruitment center, age, sex, education, Columbia MMSE/ BDRS score at initial evaluation, the Comorbidity Index, neuroleptic use, and ChEI.
Scarmeas et al. [16]	USA and Europe	456	4.5 (3.2) years	74 (8.8)	269 (59%)	21 (3.0)	Once per 6 months/9 times	NA	Cohort effect, recruitment center, informant status, sex, age, education, a comorbidity index, baseline cognitive and baseline functional performance, behavioral symptoms, and use of neuroleptics and ChEI.
Lopez et al. [37]	USA	873	1 year	75.1 (8.8)	559 (64%)	17.4 (5.1)	Once per 3 or 6 months/2–4 times	≥0.75 points/12 weeks or ≥1.5 points/24 weeks (MMSE)	Age, baseline 10-item NPI total, baseline MMSE, and diabetes.
Fong et al. [52]	Massa- chusetts	408	6 months	73.9 (8.1)	232 (56.9%)	Baseline Blessed IMC score: 11.1 (6.7)	Once per 6 months/Once	NA	Age, sex, education, comorbid medical diagnoses, family history of dementia, dementia severity score, and duration of symptoms.
Levy et al. [17]	USA	181	1 year	70.8 (7.8)	100 (55%)	20 (4)	Once per three months/4 times	NA	Age, sex
Ortof et al. [18]	USA	54	32 (12–98) months	67 (48–86)	34 (63.0%)	Blessed score: 5.4 (3–10)	Once per four months/8 times	NA	NA
Swanwick et al. [19]	Ireland	95	8–16 months	73.2 (6.4)	69 (72.6%)	19.6 (3.8)	Mean interval: 12.3 (1.4) months/1-2 times	NA	Age, education, baseline cognitive/functional scores.
Rasmusson et al. [20]	USA	132	7.5 years	70.09 (8.25)	84 (63.6%)	16.73 (3.97)	Once per 6 months/15 times	NA	Education, handedness, family history of dementia
Bruandet et al. [21]	France	670	3.0 (0.9) years	72.6 (8.0)	432 (64.5%)	20.8 (5.5)	Once per 6 to 12 months/3–6 times	NA	Age, sex, ChEIs treatment, diabetes, hypertension, visible vascular lesions on MRI, baseline DRS and MMSE.
Cortes et al. [22]	France	471	2 years	77.7 (6.77)	344 (73%)	20.16 (4.12)	Once per 6 months/4 times	NA	NA
Abellan van Kan et al. [23]	France	629	2 years	78.0 (6.75)	454 (72.2%)	20.0 (4.2)	Once per 6 months/4 times	NA	NA
Li et al. [24]	China	324	5 years	72.9 (4.14)	228 (70.4%)	15.5 (1.59)	Annually/5 times	NA	Age, gender, educational, occupation, baseline MMSE and ADL scores, and use of ChEI.
Pachalska et al. [25]	Poland	39	2 years	77.38 (8.38)	24 (61.5%)	16.08 (2.42)	Annually/2 times	NA	Vascular component
van der Vlies et al. [26]	Netherlands	291	2.3 (1.1) years	68.8 (4.90)	146 (50.2%)	21.7 (4.2)	Once per 6 months/2 times	NA	Sex, time, baseline MMSE
Deschaintre et al. [27]	France	301	2.3 years	72.0 (8.05)	69.40%	21.6 (5.0)	Once per 6 to 12 months/2–4 times	NA	Age, sex, education, first MMSE score, ChEIs, number of vascular risk factors, year of first visit, duration of symptoms before first visit, and propensity score.
Basun et al. [28]	Sweden	60	3 years	83.9 (6.3)	50 (83.3%)	13.07 (8.22)	Every 1.5 years/2 times	NA	Sex, heredity, age, and age of onset.
Kleiman et al. [29]	USA	366	1.76 years	73.6 (8.3)	223 (60.9%)	17.2 (5.5)	Once per 6 months/3.52 times	NA	Age, sex, and education.
Holmes et al. [30]	London	164	1 year	81.9 (6.6)	NA	10.7 (6.8)	Annually/Once	NA	Age at interview, age at onset, duration of illness, sex or family history of AD.
Dal Forno et al. [31]	USA	101	10 years	NA	69 (68.3%)	14.7 (5.8)	Once per 6 months/20 times	NA	Handedness, education, family history of dementia, performance on the Boston Naming Test, the Gollin Incomplete Figures Test, Block Design, the Benton Visual Retention Test, and the Responsive Naming Test.
Zhao et al. [32]	China	165	2.57 (0.99) years	74.19 (8.83)	108 (65.5%)	15.52 (5.86)	Annually/2.57 times	NA	Age, age at onset
Hirono et al. [33]	Japan	64	1 year	74.2 (6.2)	48 (75%)	19.7 (5.8)	Annually/Once	NA	Age, sex, education, test interval, and baseline scores.
Stern et al. [34]	USA	99	6 years	71.3 (7.7)	47 (47.5%)	mMMS 38.2 (5.41)	Once per 6 months/12 times	NA	Gender, age, education, ethnicity, duration of illness, age at onset, mMMS score at the initial visit.
Mortimer et al. [35]	USA	65	4 year	63.8(8.5)	NA	17.2(4.8)	Once per 6 months/8 times	NA	Age, education, sex, and initial score
Jacobs et al. [36]	USA	127	2 years	73.1 (6.01)	79 (62.2%)	mMMS 38.20 (5.82)	Once per 6 months/4 times	NA	Symptom duration, gender, family history of dementia, and baseline mMMS and BDRS score
Ousset et al. [38]	France	160	1 year	75.7 (5.8)	109 (68.1%)	23.7 (1.8)	Annually/Once	≥1 points/year (CDR)	NA
Tumini et al. [39]	Italy	321	2 years	75.3 (8.2)	66%	NA	Annually/2 times	≥5 points/year (MMSE)	Gender and age.
Silvestrini et al. [40]	Italy	411	1 year	70.9 (62–88)	227 (55.2%)	23.3 (1.9)	Once per 3 months/4 times	NA	Age, gender, diabetes, hypertension, smoking habits, education, and hypercholesterolemia.
Burns et al. [41]	London	110	1 year	80.7 (67–99)	87 (79.1%)	Mild to severe	Once per year/Once	NA	Age, age of onset, education, gender, presence of apraxia, aphasia, extrapyramidal signs, myoclonus, or parietal lobe atrophy on CT.
Huff et al. [53]	Boston	165	≥3 months	68.3 (8.8)	62%	BDS score: 24.3 (13.5)	Once per 3 months/≥Once	NA	Severity at first examination, age at first examination, estimated duration of dementia, and age at onset of dementia.
Livingston et al. [42]	London	224	≥6 months	81.0 (7.4)	160 (71.4%)	14.7 (7.6)	Once per 6 months/≥Once	NA	Demographic factors, neuropsychiatric symptoms, cognitive severity and ChEI.
Wallin et al. [43]	Sweden	151	3 years	75.8 (6.2)	104 (68.9%)	21.7 (4.4)	Once per 6 months/6 times	NA	Age, gender, APOE4 allele, and baseline MMSE level.
Kester et al. [44]	Netherlands	151	2.0 (1.0–5.0) years	67 (41–94)	68 (45%)	22 (4)	Once per 6 months/4 times	NA	Sex, age, education, and APOE genotype.
Borroni et al. [45]	Italy	48	1 year	70.7 (9.1)	34 (70.8%)	21.2 (4.80)	Once per year/Once	NA	Demographic characteristics, age at onset, or disease severity.
Soto et al. [46]	France	414	4 years	78.6 (6.9)	297 (71.7%)	16.8 (5.5)	Once per 6 months/8 times	≥3 points/6 months (MMSE)	Age, gender, education, BPSD with the total NPI score, delusions, ADAS-cog score, MNA, ADL, comorbidities, patients presenting RCD during the first year of follow- up, and the use of ChEIs.
Adams et al. [47]	Netherlands	360	6.0 (1.5) years	71.9 (7.2)	0.571	MCI	Once per 4 years/1.5 times	NA	Age and sex
Tosto et al. [48]	USA	332	50 months	74.6 (7.4)	118 (35.5%)	MCI	Once per 6 months/8 times	≥3 points/6 months or ≥6 points/year (MMSE)	Sex, age, education, and ApoE status.
Cronk et al. [49]	USA	286	1 year	75 (7.5)	NA	MCI 26.9 (1.8)	Once per 6 months/2 times	NA	Age, education, and sex
Hughes et al. [50]	USA	816	3 years	78 (7.4)	62.25%	MCI	Annually/3 times	NA	Age, gender, education, marital status, children, living arrangement, IADL dependency, depressive symptoms, overall health status, and number of prescription medications.
Kvartsberg et al. [51]	Sweden	140	4 years	67.7	64 (45.7%)	AD: 23.3 MCI: 26	Annually/4 times	NA	Age and sex

AD, Alzheimer’s disease; MCI, mild cognitive impairment; RCD, rapid cognitive decline; MMSE, Mini-Mental State Examination; ADAS-cog, Alzheimer’s Disease Assessment Scale-Cognitive; MNA, Mini Nutritional Assessment; PBSD, behavioral and psychological symptoms of dementia; ADL, Activities of Daily Living; IADL, Instrumental Activities of Daily Living; NPI, Neuropsychiatric Inventory; ChEIs, cholinesterase inhibitors; ApoE, apolipoprotein E; BDNF, brain derived neurotrophic factor; BMI, body mass index; mMMS, modified Mini-Mental State Examination; IMC, information-memory-concentration; DRS, Dementia Rating Scale; CDR, Clinical Dementia Rating; SD, standard deviation; NA, not available.

In total, the meta-analysis was comprised of 9,879 AD patients from America, Europe, and Asia. The age of participants in these studies ranged from 63.8 to 83.9 years. As for the concept of RCD, 17 studies provided the concept based on the scores of relevant scales and mostly adopted the loss of ≥3 points/year on MMSE, whereas some studies applied other criteria or just measured different decline rates in each group. Of the 35 studies, nine reported the results of both univariate and multivariate analyses.

Quality of studies included

Results of quality assessment were listed in Supplementary Table 3. The range of study quality scores included in the systematic review was from 4 to 9 points (7.65±1.11). For the meta-analysis, its score varied from 5 to 9 points (7.78±1.00). Specifically, 41.51% of studies have defects in selection aspect, 28.30% in comparability, and 52.83% in outcome.

Risk factors for RCD in AD patients

Thirteen risk factors were reported to be associated with RCD in AD, which were generally classified into unmodifiable risk factors and modifiable risk factors (Supplementary Tables 4 and 5). Furthermore, the forest plot of risk factors was listed in Supplementary Figures 1 to 25.

Unmodifiable risk factors

As for the influence of age, the pooled analysis of five studies (n = 1,538) indicated that AD patients with older age (≥75 years) (univariate: RR = 1.01, 95% CI = 0.64 to 1.59, I² = 0%, n = 358; multivariate: RR = 0.96, 95% CI = 0.93 to 0.99, I² = 0%, n = 1,334) showed slower cognitive decline than younger patients (Fig. 2). Six studies (n = 1,824) that reported associations with gender or severity of global cognitive impairment were included. Impacts were identified for neither male (univariate: RR = 0.93, 95% CI = 0.41 to 2.14, I² = 65% and n = 310; multivariate: RR = 0.95, 95% CI = 0.53 to 1.70, I² = 71%, n = 747) (Supplementary Figure 1) nor moderate dementia (univariate: RR = 1.23, 95% CI = 0.51 to 2.97, I² = 60%, n = 358; multivariate: RR = 1.65, 95% CI = 0.76 to 3.59, I² = 91%, n = 1,183) (Supplementary Figure 2) and severe dementia (univariate: RR = 1.00, 95% CI = 0.31 to 3.21, n = 154; multivariate: RR = 1.55, 95% CI = 0.11 to 21.05, I² = 95%, n = 1,027) (Supplementary Figure 3). As for age at disease onset, results showed that cognitive function in early-onset AD patients deteriorated more rapidly than late-onset patients in global cognition (SMD [95% CI]: –0.65 [–0.83, –0.47], I² = 46%, n = 583) (Fig. 3). Ten studies (n = 2,236) evaluated the association of ApoE4 with RCD. Compared with non-carriers, the rate of decline in MMSE scores was increased in ApoE4 carriers (SMD [95% CI]: 0.52 [0.06, 0.98], I² = 91%, n = 1,178) (Fig. 4). By contrast, family history of dementia showed no significant result (univariate: RR = 1.11, 95% CI = 0.73 to 1.71, I² = 47%, n = 365; multivariate: RR = 1.20, 95% CI = 0.73 to 1.97, I² = 33%, n = 365) (Supplementary Figure 4).

Fig. 2

Forest plot comparing cognitive decline in older AD patients with that in younger patients (binary data). In AD patients, older age showed no significant association with cognitive decline in univariate analysis, whereas in multivariate analysis it rendered 1% to 7% decrease risk for cognitive decline.

Fig. 3

Forest plot comparing cognitive decline in early-onset AD patients with that in late-onset AD patients (continuous data). In AD patients, early-onset AD rendered 47% to 83% increase risk for cognitive decline.

Fig. 4

Forest plot comparing cognitive decline in AD patients having ApoE4 with those not having ApoE4 (continuous data). In AD patients, ApoE4 rendered 6% to 98% increase risk for cognitive decline.

Modifiable risk factors

Lifestyle

Five studies (n = 936) investigated the effect of lifestyle on the progression of cognitive decline. As for education, highly educated patients (univariate: RR = 1.50, 95% CI = 1.04 to 2.17, I² = 36%, n = 856; multivariate: RR = 2.05, 95% CI = 1.26 to 3.33, I² = 14%, n = 514) (Fig. 5) exhibited a faster trend of cognitive progression. While smoking showed no significant correlation with the rate of cognitive deterioration (univariate: RR = 1.10, 95% CI = 0.50 to 2.41, n = 156; multivariate: RR = 0.91, 95% CI = 0.36 to 2.34, I² = 9%, n = 236) (Supplementary Figure 5).

Fig. 5

Forest plot comparing cognitive decline in AD patients having high level of education with those not having high level of education (binary data). In AD patients, a high level of education increased the risk of cognitive decline by 0.04 to 1.17 times in univariate analysis, and in multivariate analysis, it increased the risk of cognitive decline by 0.26 to 2.33 times.

Fig. 6

Forest plot comparing cognitive decline in AD patients having extrapyramidal signs with those not having extrapyramidal signs. In AD patients, extrapyramidal signs showed no significant association with cognitive decline in univariate analysis, whereas in multivariate analysis, it increased the risk of cognitive decline by 0.30 to 2.67 times.

Symptoms and signs of AD

Five studies (n = 1,167) compared the rate of cognitive decline between patients with and without extrapyramidal signs and neuropsychiatric conditions (hallucination, agitation, psychosis, and strolling). For subjects with extrapyramidal signs, the rate of cognitive decline was higher than those without extrapyramidal signs (univariate: RR = 0.97, 95% CI = 0.35 to 2.71, n = 204; multivariate: RR = 2.18, 95% CI = 1.30 to 3.67, I² = 0%, n = 371) (Fig. 6). Four neuropsychiatric conditions were discovered as predictors of rapid decliners, including hallucination (univariate: RR = 2.01, 95% CI = 1.40 to 2.87, I² = 0%, n = 660; multivariate: RR = 2.17, 95% CI = 0.97 to 4.86, I² = 56%, n = 591), strolling (univariate: RR = 1.99, 95% CI = 1.38 to 2.86, I² = 0%, n = 660), agitation (univariate: RR = 1.66, 95% CI = 1.23 to 2.24, I² = 0%, n = 701; multivariate: RR = 1.98, 95% CI = 1.15 to 3.41, I² = 45%, n = 632), and psychosis (multivariate: RR = 1.41, 95% CI = 1.04 to 1.91, I² = 0%, n = 371) (Fig. 7).

Fig. 7

Forest plot comparing cognitive decline in AD patients having neuropsychiatric conditions with those not having neuropsychiatric conditions. In AD patients, hallucination increased the risk of cognitive decline by 0.40 to 1.87 times in univariate analysis, whereas in multivariate analysis, it showed no significant association with cognitive decline. Strolling increased the risk of cognitive decline by 0.38 to 1.86 times in univariate analysis. Agitation increased the risk of cognitive decline by 0.23 to 1.24 times in univariate analysis, and in multivariate analysis, it increased the risk of cognitive decline by 0.15 to 2.41 times. Psychosis increased the risk of cognitive decline by 0.04 to 0.91 times in multivariate analysis.

Medications

As for medications, multidrug therapy (multivariate: RR = 0.61, 95% CI = 0.60 to 0.62, I² = 0%, n = 403) can slow down cognitive deterioration (Fig. 8), whereas no significant difference in the rate of cognitive decline can be found in patients using neuropsychiatric drugs (univariate: RR = 1.70, 95% CI = 0.89 to 3.26, n = 154; multivariate: RR = 1.01, 95% CI = 0.61 to 1.68, I² = 63%, n = 1,222) (Supplementary Figure 6) and ChEIs (univariate: RR = 1.91, 95% CI = 0.69 to 5.31, I² = 73%, n = 360; multivariate: RR = 0.99, 95% CI = 0.55 to 1.78, I² = 71%, n = 1,958) (Supplementary Figure 7).

Fig. 8

Forest plot comparing cognitive decline in AD patients having medications with those not having medications (binary data). In AD patients, medications rendered 38% to 40% decrease risk for cognitive decline in multivariate analysis.

Comorbidities

Eight studies (n = 2,322) investigated the effects of several comorbidities on RCD. Patients with diabetes were identified to have slower cognitive decline (univariate: RR = 0.50, 95% CI = 0.27 to 0.93, I² = 39%, n = 310; multivariate: RR = 0.58, 95% CI = 0.36 to 0.94, I² = 0%, n = 732) (Fig. 9). However, other comorbidities showed no significant correlation with RCD, including hypertension, hypercholesterolemia, heart disease, and other vascular risk factors (Supplementary Figures 8–11). Furthermore, no correlation was found between the rate of cognitive progression and activities of daily living (ADL) score (multivariate: RR = 0.65, 95% CI = 0.30 to 1.40, I² = 95%, n = 1,066) (Supplementary Figure 12).

Fig. 9

Forest plot comparing cognitive decline in AD patients having diabetes with those not having diabetes (binary data). In AD patients, comorbidity diabetes rendered 7% to 73% decrease risk for cognitive decline in univariate analysis, and in multivariate analysis, it rendered 6% to 64% decrease risk for cognitive decline.

Biochemical factors

The risk of RCD increased in patients with higher levels of CSF T-tau (multivariate: RR = 3.31, 95% CI = 1.53 to 7.16, n = 196) or CSF P-tau (multivariate: RR = 2.53, 95% CI = 1.21 to 5.26, n = 196) according to a retrospective cohort study (Supplementary Table 5). Our result showed high levels of P-tau are not associated with cognitive decline in AD patients via pooling related continuous data (SMD [95% CI]: –0.42 [–1.06, 0.22], I² = 61%, n = 302) (Supplementary Figure 13). Additionally, elevated total homocysteine (tHcy) levels in plasma were also estimated as an independent risk factor for RCD (multivariate: RR = 1.74, 95% CI = 1.16 to 2.60, n = 80), especially for executive function (Supplementary Table 6).

Risk factors of RCD in MCI patients

Five studies were included in a systematic review to illustrate the association between relevant factors and the progression of cognitive decline in MCI. Higher BMI, and greater engagement in social activities were reported as predictive factors of slower cognitive decline, while other five factors were associated with an increased risk of cognitive deterioration, including ApoE4 status, greater white matter hyperintensities (WMHs) volume, smaller entorhinal cortex volume (ECV), high CSF neurogranin (Ng) levels, and high instrumental activities of daily living (IADL) dependency (Supplementary Table 7). Additionally, the pooled outcome of risk factors for RCD was listed in Supplementary Table 8.

Sensitivity analyses

Evident heterogeneity was found in thirteen analyses (I² >50%, p < 0.05). After sensitivity analyses, the estimated effects of these factors barely changed except for moderate dementia and ApoE4. Specifically, for the association between moderate dementia and cognitive progression, the heterogeneity was reduced to 0% with the effect size showing positive association for cognitive deterioration (RR = 2.49, 95% CI = 1.78 to 3.49) after one study was removed which performed only one assessment of cognitive function in the 18-month follow-up [17]. As for ApoE4, adjusting heterogeneity (I² = 31%) leads to no significant effect (SMD [95% CI]: –0.04[–0.15, 0.06]) when two studies neither of which provided a definite concept of RCD were excluded [18, 19], suggesting that the potential role of ApoE4 in RCD requires more high-quality studies to prove.

DISCUSSION

In this systematic review and meta-analysis, seventeen factors were indicated to be potential predictors of rapid cognitive decline in AD and MCI patients. For patients with AD, ApoE4, high level of education, early age at onset, early appearance of extrapyramidal signs, and psychiatric conditions predicted a faster rate of cognitive deterioration. Conversely, older age, diabetes, and multidrug therapy would slow the rate of cognitive decline. Additionally, for other factors listed in Supplementary Tables 4–6, the evidence is less persuasive due to sample size limitations and quality issues, and thus more studies are required to confirm their relationship with cognitive progression.

Initially in our research, no relation was found between disease severity and cognitive progression. After sensitivity analyses were conducted, moderate cognitive impairment (16<MMSE ≤21) was revealed as a significant risk factor for RCD, but this remarkable association could not be found in severe cognitive impairment (MMSE ≤16) patients. Some studies owed this to the less sensitivity of MMSE for brain damage in the early and late stages of the disease [20]. Others proposed that it is in accordance with neurodegenerative changes in AD patients, and suggested the slow rate of cognitive decline in severe dementia probably results from little remaining cognitive capacity to be damaged [21]. Another explanation for this is the presence of “floor effect” in severe stage of AD. They thought that the progressive rate of cognitive function was associated with cerebral atrophy and severe stage AD individuals had already reached a maximum degree of cerebral atrophy, and subsequently reflecting a relative slow progression of cognitive capacity [22].

ApoE4 is regarded as the strongest genetic predictor of the development of sporadic AD [4]. Its positive correlation with RCD in AD patients was also revealed in our meta-analysis. One reasonable explanation for this is that ApoE4 allele carriers would disturb neurofibrillary tangle biochemical pathways and have an impact on amyloid-β (Aβ) accumulation [23]. It is also proposed that ApoE4 allele is associated with rapid progression of hippocampal atrophy in AD patients according to a longitudinal MRI volumetric study [24], subsequently resulting in more rapid cognitive deterioration. However, the difference became no statistically significant after conducting sensitivity analysis via excluding two studies which did not provide a definite concept of RCD. This is consistent with those of a meta-analysis conducted by Allan et al. [25] and a longitudinal cohort study conducted by Hoyt et al. [26]. The above results indicated that the potential role of ApoE4 in RCD is ambiguous, which is resulting from the definition of RCD. Additionally, the discrepancy in the influence of ApoE4 on cognitive progression in AD patients may be ascribed to different samples included in each study, which explained that in moderate and severe stages of disease or after some neurobiological threshold was crossed, more advanced illness appeared and limited our ability to detect ApoE4-associated deterioration [23, 25]. As for polygenic risk scores (PRS) which include ApoE4, it was reported to accelerate the cognitive decline (executive function, memory, and CDR-SB score) in AD [27, 28] and MCI [29] patients by the linear mixed effects models. However, PRS was hardly related to the cognitive complaints in MCI patients by the logistic regression models [30]. This did not correspond with its effect on MCI to AD dementia progression [31]. Related studies provided both cognitive decline and time to AD progression resulting from the combination of PRS, amyloid, and total tau but not amyloid or total tau alone [29].

Education was identified as a risk factor for the progression of cognitive decline in AD. It was previously reported that high level of education had a protective impact on the incidence of dementia but would trigger a faster cognitive decline once clinical symptoms appear [32]. Some researchers suggested high educated patients would experience a faster decline in cognition and most of them explained this using the “cognitive reserve” hypothesis. This hypothesis predicted that persons with high education would have advanced cognitive reserve, and thus they would be more resistant to the impact of AD pathology, leading to delayed manifestation of clinical dementia. However, due to their advanced reserve, high educated individuals subsequently showed rapid cognitive decline once clinical symptoms emerged [1]. Some also proposed that highly educated persons could escape early detection by scoring well on cognitive screening scales, leading to delayed performance of observable dysfunction in the process of disease, at which point accelerated cognitive decline begins to occur [33]. Therefore, the inconsistency about faster rate of cognitive deterioration observed in less educated person might just reflect the impact of early AD neuropathology.

In addition, neither smoking nor past smoking [34] showed significant association with the rate of cognitive decline in AD. However, researchers found AD patients with smoking ≥1 packet of cigarette every day had fast cognitive decline after controlling the confounding factors [17]. It is consistent with the previous finding that current smokers had faster decline in the MMSE score [35]. Further, it was also previously shown that smoking would reduce the volume of basal forebrain but not hippocampus in healthy individuals and early MCI, whereas this association could not be found in late MCI and AD dementia [36]. Possible mechanisms about the correlation between smoking and RCD include impact of small vessel disease on the brain in smokers, volume change of WMH, or adverse effects of chronic nicotine on cholinergic system [36 –38].

Extrapyramidal signs and neuropsychiatric conditions in AD patients can predict a faster cognitive deterioration. Their effects on RCD might be related to pathological involvement of brain structures or alterations of regional cerebral metabolism [39, 40]. Furthermore, many specific neuropsychiatric conditions, such as agitation and hallucination, were also identified as risk factors for cognitive decline. However, as for the specific extrapyramidal signs, although many studies have reported the association of related motor function with the risk of incident dementia [41], there are few studies to identify its relation with RCD, so further research is needed to explore this relationship. Neuropsychiatric conditions could be managed by relevant medications, psychotherapy, and so on [42, 43], but it is unclear whether these managements could play a role in slowing disease progression. Further studies should be conducted to clarify the association between antipsychotics and cognitive deterioration.

In this study, AD patients with diabetes have slower progression rate of cognitive decline. Some studies suggested the superficial protective effect of diabetes was probably due to the effects of antidiabetic medications or insulin degrading enzyme (IDE) on the formation of Aβ peptides and amyloid plaque. Notably the latter was also related to insulin [44]. It was proposed that the paradox may be due to the different stages of diabetes [44]. Specifically, IDE is inadequate for the degradation of Aβ peptides and insulin in the early stage of type 2 diabetes, which could result in abnormal augment of Aβ accumulation. Besides, insulin would increase the production of insulin itself, which could enhance the competition of insulin for IDE. However, when the insulin levels decreased owing to failure of islet cells in diabetes patients, the degradation of Aβ peptides is active in contrast to their non-diabetic counterparts. The mechanisms of two different stages may explain the protective role of diabetes in the progression of cognitive decline in AD, which could also explain the different results may be due to the different follow-up periods. Furthermore, the differences between positive and neutral or negative studies were also compared, one of which is the longer follow period in positive studies [11].

This is the first systematic review and meta-analysis of the risk factors for RCD using both univariate and multivariate analyses, which could clearly show the effects of confounding factors on the results. The strengths include a comprehensive literature search, duplicate study screening and data extraction, and credibility of risk factors guaranteed by restricting case number, quality of included studies as well as the heterogeneity of the studies. However, there are many limitations in the present study. First, a consensus definition of RCD has not yet been developed. In our research, some studies used the drop in the MMSE to distinguish between rapid decliners and slow decliners, whereas others applied different criterion or just measured different decline rates in each group. Although the majority of the researchers measured the progression of cognitive decline by MMSE score, some weaknesses still existed, including its poor sensitivity to detect brain changes in early and late stages of disease as well as the presence of “floor effect” in severe stages [45, 46]. Many researchers found that the significant effect of moderate dementia on the progression of cognitive decline was ascribed to this limitation of MMSE score [47]. Furthermore, the episode of RCD was probably regarded as a dynamic condition and individuals suffering it were divided into two types: the “primary” rapid cognitive decliners and rapid decliners due to a specific reason [1]. But both types of patients showed a similar worse prognosis related to mortality and autonomy loss. This heterogeneity in RCD definition may influence the accuracy of final results. Considering this condition, we only chose the studies reporting MMSE or other scale scores definitely. Second, our research only included the articles published in English, which may lead to the missing of relevant studies. Third, we failed to conduct more detailed analyses for some risk factors because the number of these studies was relatively small and not enough to do the subgroup analyses. Similarly, we did not create funnel plots to test the publication bias due to the limitation of the number of studies included for each risk factor.

In conclusion, this systematic review and meta-analysis emphasized ApoE4, age, and age at disease onset as unmodifiable risk factors that can predict the rapid and aggressive AD course. Furthermore, education, early appearance of neuropsychiatric conditions and extrapyramidal signs, pre-existing cognitive impairment, diabetes, and medications as modifiable risk factors might be potentially effective targets for disease prediction and early intervention. All these findings may provide new insights into developing predictive models and therapeutic approaches to patient management. Further high-quality studies with longer follow-up periods and larger sample sizes are needed to confirm our results.

Footnotes

ACKNOWLEDGMENTS

This work was supported by grants from the National Key R&D Program of China (2016YFC1305803), the National Natural Science Foundation of China (81771148, 81471309, 81571245, 81501103, and 81701253), the Shandong Provincial Outstanding Medical Academic Professional Program, Taishan Scholars Program of Shandong Province (ts201511109, tsqn20161078 and tsqn20161079), Qingdao Key Health Discipline Development Fund, Qingdao Outstanding Health Professional Development Fund, and Shandong Provincial Collaborative Innovation Center for Neurodegenerative Disorders.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Soto

, Andrieu

, Arbus

, Ceccaldi

, Couratier

, Dantoine

, Dartigues

, Gillette-Guyonnet

, Nourhashemi

, Ousset

, Poncet

, Portet

, Touchon

, Vellas

(2008) Rapid cognitive decline in Alzheimer’s disease. Consensus paper. JNutr Health Aging 12, 703–713.

(2016) 2016 Alzheimer’s disease facts and figures. Alzheimers Dement 12, 459–509.

Petersen

, Caracciolo

, Brayne

, Gauthier

, Jelic

, Fratiglioni

(2014) Mild cognitive impairment: A concept in evolution. J Intern Med 275, 214–228.

, Tan

, Wang

, Tan

, Xu

, Zhao

, Wang

, Jiang

, Yu

(2016) Risk factors for predicting progression from mild cognitive impairment to Alzheimer’s disease: A systematic review and meta-analysis of cohort studies. J Neurol Neurosurg Psychiatry 87, 476–484.

Carcaillon

, Peres

, Pere

, Helmer

, Orgogozo

, Dartigues

(2007) Fast cognitive decline at the time of dementia diagnosis: A major prognostic factor for survival in the community. Dement Geriatr Cogn Disord 23, 439–445.

Hui

, Wilson

, Bennett

, Bienias

, Gilley

, Evans

(2003) Rate of cognitive decline and mortality in Alzheimer’s disease. Neurology 61, 1356–1361.

Holtzer

, Wegesin

, Albert

, Marder

, Bell

, Albert

, Brandt

, Stern

(2003) The rate of cognitive decline and risk of reaching clinical milestones in Alzheimer disease. ArchNeurol 60, 1137–1142.

Wilson

, Li

, Aggarwal

, McCann

, Gilley

, Bienias

, Barnes

, Evans

(2006) Cognitive decline and survival in Alzheimer’s disease. IntJ Geriatr Psychiatry 21, 356–362.

Jia

, Gauthier

, Pallotta

, Ji

, Wei

, Xiao

, Peng

, Guo

, Wu

, Chen

, Kuang

, Zhang

, Wei

, Tang

(2017) Consensus-based recommendations for the management of rapid cognitive decline due to Alzheimer’s disease. Alzheimers Dement 13, 592–597.

10.

Sona

, Ellis

, Ames

(2013) Rapid cognitive decline in Alzheimer’s disease: A literature review. Int Rev Psychiatry 25, 650–658.

11.

, Cesari

, Liu

, Dong

, Vellas

(2017) Effects of diabetes mellitus on cognitive decline in patients with Alzheimer disease: A systematic review. Can J Diabetes 41, 114–119.

12.

Stang

(2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 25, 603–605.

13.

Wang

, Tan

, Wang

, Tan

, Meng

, Wang

, Tang

, Yu

(2015) Anti-inflammatory drugs and risk of Alzheimer’s disease: An updated systematic review and meta-analysis. J Alzheimers Dis 44, 385–396.

14.

Higgins

, Thompson

, Deeks

, Altman

(2003) Measuring inconsistency in meta-analyses. BMJ 327, 557–560.

15.

DerSimonian

, Laird

(2015) Meta-analysis in clinical trials revisited. Contemp Clin Trials 45, 139–145.

16.

Duval

, Tweedie

(2000) Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 56, 455–463.

17.

Sona

, Zhang

, Ames

, Bush

, Lautenschlager

, Martins

, Masters

, Rowe

, Szoeke

, Taddei

, Ellis

(2012) Predictors of rapid cognitive decline in Alzheimer’s disease: Results from the Australian imaging, biomarkers and lifestyle (AIBL) study of ageing. Int Psychogeriatr 24, 197–204.

18.

Dal Forno

, Rasmusson

, Brandt

, Carson

, Brookmeyer

, Troncoso

, Kawas

(1996) Apolipoprotein E genotype and rate of decline in probable Alzheimer’s disease. Arch Neurol 53, 345–350.

19.

Basun

, Grut

, Winblad

, Lannfelt

(1995) Apolipoprotein epsilon 4 allele and disease progression in patients with late-onset Alzheimer’s disease. Neurosci Lett 183, 32–34.

20.

Musicco

, Salamone

, Caltagirone

, Cravello

, Fadda

, Lupo

, Mosti

, Perri

, Palmer

(2010) Neuropsychological predictors of rapidly progressing patients with Alzheimer’s disease. Dement Geriatr Cogn Disord 30, 219–228.

21.

Stern

, Mohs

, Davidson

, Schmeidler

, Silverman

, Kramer-Ginsberg

, Searcey

, Bierer

, Davis

(1994) A longitudinal study of Alzheimer’s disease: Measurement, rate, and predictors of cognitive deterioration. Am J Psychiatry 151, 390–396.

22.

Burns

, Jacoby

, Levy

(1991) Progression of cognitive impairment in Alzheimer’s disease. J Am Geriatr Soc 39, 39–45.

23.

Cosentino

, Scarmeas

, Helzner

, Glymour

, Brandt

, Albert

, Blacker

, Stern

(2008) APOE epsilon 4 allele predicts faster cognitive decline in mild Alzheimer disease. Neurology 70, 1842–1849.

24.

Hirono

, Hashimoto

, Yasuda

, Kazui

, Mori

(2003) Accelerated memory decline in Alzheimer’s disease with apolipoprotein epsilon4 allele. J Neuropsychiatry Clin Neurosci 15, 354–358.

25.

Allan

, Ebmeier

(2011) The influence of ApoE4 on clinical progression of dementia: A meta-analysis. Int J Geriatr Psychiatry 26, 520–526.

26.

Hoyt

, Massman

, Schatschneider

, Cooke

, Doody

(2005) Individual growth curve analysis of APOE epsilon 4-associated cognitive decline in Alzheimer disease. Arch Neurol 62, 454–459.

27.

Desikan

, Fan

, Wang

, Schork

, Cabral

, Cup-ples

, Thompson

, Besser

, Kukull

, Holland

, Chen

, Brewer

, Karow

, Kauppi

, Witoe-lar

, Karch

, Bonham

, Yokoyama

, Rosen

, Miller

, Dillon

, Wilson

, Hess

, Pericak-Vance

, Haines

, Farrer

, Mayeux

, Hardy

, Goate

, Hyman

, Schellenberg

, McEvoy

, Andreassen

, Dale

(2017) Genetic assessment of age-associated Alzheimer disease risk: Development and validation of a polygenic hazard score. PLoS Med 14, e1002258.

28.

Louwersheimer

, Espinosa

, Wolfsgruber

, Lacour

, Heilmann-Heimbach

, Hernandez

, Holstege

, Fernandez

, Wagner

, Rosende-Roca

, Mauleon

, Moreno-Grau

, Vargas

, Pijnenburg

YAL

, Koene

, Rodriguez-Gomez

, Ortega

, Ruiz

, Sotolongo-Grau

, Kornhuber

, Peters

, Frolich

, Hull

, Ruther

, Wiltfang

, Scherer

, Riedel-Heller

, Alegret

, Nothen

, Scheltens

, Tarraga

, Jessen

, Boada

, Maier

, Becker

, Wagner

, Ruiz

, Van Der Flier

, Ramirez

(2016) Role of Alzheimer’s disease risk genes in the progression to dementia and in relevant endophenotypes among subjects with mild cognitive impairment. Med Genet 28, 178.

29.

Tan

, Fan

, Mormino

, Sugrue

, Broce

, Hess

, Dillon

, Bonham

, Yokoyama

, Karch

, Brewer

, Rabinovici

, Miller

, Schellenberg

, Kauppi

, Feldman

, Holland

, McEvoy

, Hyman

, Bennett

, Andreassen

, Dale

, Desikan

; Alzheimer’s Disease Neuroimaging Initiative (2018) Poly-genic hazard score: An enrichment marker for Alzheimer’s associated amyloid and tau deposition. Acta Neuropathol 135, 85–93.

30.

Adams

, de Bruijn

, Hofman

, Uitterlinden

, van Duijn

, Vernooij

, Koudstaal

, Ikram

(2015) Genetic risk of neurodegenerative diseases is associated with mild cognitive impairment and conversion to dementia. Alzheimers Dement 11, 1277–1285.

31.

Lacour

, Espinosa

, Louwersheimer

, Heilmann

, Hernandez

, Wolfsgruber

, Fernandez

, Wagner

, Rosende-Roca

, Mauleon

, Moreno-Grau

, Vargas

, Pijnenburg

, Koene

, Rodriguez-Gomez

, Ortega

, Ruiz

, Holstege

, Sotolongo-Grau

, Kornhuber

, Peters

, Frolich

, Hull

, Ruther

, Wiltfang

, Scherer

, Riedel-Heller

, Alegret

, Nothen

, Scheltens

, Wagner

, Tarraga

, Jessen

, Boada

, Maier

, van der Flier

, Becker

, Ramirez

, Ruiz

(2017) Genome-wide significant risk factors for Alzheimer’s disease: Role in progression to dementia due to Alzheimer’s disease among subjects with mild cognitive impairment. Mol Psychiatry 22, 153–160.

32.

Meng

, D'Arcy

(2012) Education and dementia in the context of the cognitive reserve hypothesis: A systematic review with meta-analyses and qualitative analyses. PLoS One 7, e38268.

33.

Teri

, McCurry

, Edland

, Kukull

, Larson

(1995) Cognitive decline in Alzheimer’s disease: A longitudinal investigation of risk factors for accelerated decline. J Gerontol A Biol Sci Med Sci 50a, M49–55.

34.

Bhargava

, Weiner

, Hynan

, Diaz-Arrastia

, Lip-ton

(2006) Vascular disease and risk factors, rate of progression, and survival in Alzheimer’s disease. J Geriatr Psychiatry Neurol 19, 78–82.

35.

Anstey

, von Sanden

, Salim

, O'Kearney

(2007) Smoking as a risk factor for dementia and cognitive decline: A meta-analysis of prospective studies. Am J Epidemiol 166, 367–378.

36.

Teipel

, Grothe

; Alzheimer’s Disease Neuroimaging Initiative (2016) Association between smoking and cholin-ergic basal forebrain volume in healthy aging and prodromal and dementia stages of Alzheimer’s disease. J Alzheimers Dis 52, 1443–1451.

37.

Staals

, Makin

, Doubal

, Dennis

, Wardlaw

(2014) Stroke subtype, vascular risk factors, and total MRI brain small-vessel disease burden. Neurology 83, 1228–1234.

38.

Dickie

, Ritchie

, Cox

, Sakka

, Royle

, Aribisala

, Valdes Hernandez Mdel

, Maniega

, Pattie

, Corley

, Starr

, Bastin

, Deary

, Wardlaw

(2016) Vascular risk factors and progression of white matter hyperintensities in the Lothian Birth Cohort 1936. Neurobiol Aging 42, 116–123.

39.

Chui

, Lyness

, Sobel

, Schneider

(1994) Extrapyramidal signs and psychiatric symptoms predict faster cognitive decline in Alzheimer’s disease. Arch Neurol 51, 676–681.

40.

Murray

, Kumar

, Demichele-Sweet

, Sweet

(2014) Psychosis in Alzheimer’s disease. Biol Psychiatry 75, 542–552.

41.

Kueper

, Speechley

, Lingum

, Montero-Odasso M (2017) Motor function and incident dementia: A systematic review and meta-analysis. Age Ageing 46, 729–738.

42.

Wang

, Yu

, Wang

, Meng

, Wang

, Tan

(2015) Pharmacological treatment of neuropsychi-atric symptoms in Alzheimer’s disease: A systematic review and meta-analysis. J Neurol Neurosurg Psychiatry 86, 101–109.

43.

Lawlor

(2004) Behavioral and psychological symptoms in dementia: The role of atypical antipsychotics. J Clin Psychiatry 65(Suppl 11), 5–10.

44.

Musicco

, Palmer

, Salamone

, Lupo

, Perri

, Mosti

, Spalletta

, di Iulio

, Pettenati

, Cravello

, Caltagirone

(2009) Predictors of progression of cognitive decline in Alzheimer’s disease: The role of vascular and sociodemo-graphic factors. J Neurol 256, 1288–1295.

45.

Soto

, Andrieu

, Cantet

, Reynish

, Ousset

, Arbus

, Gillette-Guyonnet

, Nourhashemi

, Vellas

(2008) Predictive value of rapid decline in mini mental state examination in clinical practice for prognosis in Alzheimer’s disease. Dement Geriatr Cogn Disord 26, 109–116.

46.

Flicker

(2010) Screening and assessment instruments for the detection and measurement of cognitive impairment.

47.

Ito

, Corrigan

, Zhao

, French

, Miller

, Soares

, Katz

, Nicholas

, Billing

, Anziano

, Fullerton

(2011) Disease progression model for cognitive deterioration from Alzheimer’s Disease Neuroimaging Initiative database. Alzheimers Dement 7, 151–160.