Progression of Alzheimer’s Disease: A Longitudinal Study in Norwegian Memory Clinics

INTRODUCTION

Alzheimer’s disease (AD) is the most common cause of dementia, accounting for 50–70% of cases. Because of population aging, AD affects increasing numbers of people worldwide. The progressive disorder leads to increasing disability, with substantial impact on individuals, families, and societies. Therefore, to decrease the disease burden, it is highly important both to reduce the risk of developing AD and to slow its progression.

Symptoms and impairments resulting from AD usually become apparent at the prodromal stage of dementia, also called mild cognitive impairment (MCI) [1]. In addition to the cognitive decline, disease progression in AD leads to a loss of functional abilities, typically with initial impairments in instrumental activities of daily living (IADL) and later impairments in basic ADL. Thus, disease progression can be measured with scales that assess cognition or function, or with global scales that provide a compound measure of disease severity. The cognitive and functional decline may be disparate and describing the changes in more than one way may give a more thorough picture of the situation [2].

As a global assessment scale measuring progression of AD, the Clinical Dementia Rating (CDR), with its continuous scale the Clinical Dementia Rating Sum of Boxes (CDR-SB), is frequently used. A mean yearly change in CDR-SB between 1.4 and 2.3 points is reported for patients with AD. Disease progression in AD is generally described as slower in the initial stages than in a more advanced stage of the disease, with yearly progression in CDR-SB of 0.5 to 1.9 points among patients with amnestic MCI (aMCI) or very mild dementia (CDR 0.5) and 1.3 to 1.9 points among patients with mild dementia (CDR 1) [3–6]. There is considerable individual variation in progression rates [7, 8]. Using the Mini-Mental State Examination (MMSE) to measure cognition, the average loss is between 1.5 and 2.3 points per year [5, 9]. A rapid cognitive decline, defined as a decrease of 3 to 5 points per year on the MMSE [8], has been shown in 10–30% of AD cases, depending on the threshold set and the population studied [8, 10]. Few studies have examined factors related to slow decline in AD [9].

A population-based study found that 30–58% of the participants with AD declined slowly, with a yearly change of less than one point on the CDR-SB and the MMSE [6]. Consistent with this finding, two memory clinic studies found that 25–30% of the patients were stable over a period of two to three years [5, 8].

Some studies have shown that young age at onset, high educational level, and severity of cognitive and functional impairment at the time of diagnosis are associated with rapid decline in AD patients [11]. There is still a considerable lack of evidence on other predictors for the rate of decline in individual patients with AD [12]. Knowledge about the prognosis for individual patients is very important to enable better planning for the future. Further, the identification of modifiable predictors for the disease course could be a target for future intervention studies focusing on treatment that might hamper disease progression. Therefore, more research is needed on well-defined and clinical AD populations focusing on various aspects of disease progression.

The primary aim of the present study was to study the overall progression of AD, as measured by the primary outcome measure CDR-SB. Further, we wanted to investigate whether baseline characteristics are of importance for differences in progression, and to examine the correlation between the progression assessed by a global score (CDR-SB) and progression in cognitive (MMSE) and functional (IADL) measures.

MATERIALS AND METHODS

RESULTS

Of the 555 patients with cognitive impairment or dementia included at baseline, 198 were lost to follow-up, and 75 had other causes of cognitive impairment than AD (Fig. 1). Patients lost to follow-up had almost one year less of formal education, were more likely to live alone (38% versus 28%) and to be smoking (20% versus 11%), as compared to patients who remained in the study. A higher proportion of patients lost to follow-up had a score below – 2 SD on the Trail Making Test A (46% versus 36%) and B (69% versus 52%). There were no significant group differences for other cognitive tests nor for comorbidities.

OPEN IN VIEWER

Fig.1

Study flow chart. Other types of dementia (n = 40) among patients with follow-up assessments: Dementia in Parkinson’s disease (n = 12). Dementia with Lewy bodies (n = 8). Vascular dementia (n = 7). Frontotemporal dementia (n = 4). In addition, two patients were diagnosed with other specified types of dementia and seven had unspecified dementia.

Of the 282 AD patients who participated in follow-up, 155 had a global CDR score of 0.5 (very mild AD), 105 had a CDR score of 1 (mild AD), and 22 had a CDR score of 2 (moderate AD) at baseline. Of the 105 aMCI patients, 98 had a CDR score of 0.5; the rest had CDR 1. The baseline characteristics of the AD patients are summarized in Table 1. The mean age was 73.3±8.8 years, mean length of education 11.7±3.6 years, and 54.3% were females. The mean MMSE score was 23.7±4.4, mean IADL score 0.78±0.22 and mean CDR-SB score 4.2±2.8. Symptoms started before age 65 in 19% of patients with dementia and in 38% of patients with MCI at baseline. Half of the aMCI patients with symptoms before age 65 had developed dementia by follow-up.

As shown in Table 2, the mean increase (worsening) in CDR-SB was 1.6±1.8 points/year (cut-points for highest and lowest quartiles 0.3 and 2.2 points per year). We found no differences in progression between users (n = 122) and non-users (n = 160) of cholinesterase inhibitors or memantine.

Table 3 shows the results of linear regression analyses for predictors associated with CDR-SB change. In the unadjusted linear regression analyses, there was a statistically significant association between age, number of drugs taken regularly, Trail Making Test B scores, the CERAD word-list delayed recall, and change in CDR-SB. In the adjusted analysis, the Trail Making Test B scores remained statistically significant. A model including age, ApoE ɛ4 carrier status, Trail Making Test B score, word-list delayed recall score, history of hypertension, and the number of drugs taken regularly explained R² = 17% of the variance of the change in CDR-SB score. With a mean variance inflation factor of 1.19 and tolerance statistics between 0.76 and 0.91, multicollinearity was not found to be a problem for this model.

The number of slow progressors in terms of CDR-SB change was 132 (46.8%), with 32 patients (11.3%) having better CDR-SB score at follow-up, 20 (7.1%) unchanged score, and 80 (28.4%) with less than 1 point increase in CDR-SB per year. The number of patients who were slow progressors was 80 (51.6%) among patients with very mild AD, 46 (43.8%) among patients with mild AD and six (27.3%) among those with moderate AD (p = 0.075). As shown in Table 4, slow progressors were younger at diagnosis, had longer formal education, used fewer drugs, had slightly better IADL function, better scores on cognitive tests, and scored one point less on CDR-SB at baseline. There were no differences in the use of anti-dementia medications or in depressive symptoms among individuals with improved scores as compared to the rest of the patients.

The mean decline in MMSE scores was 1.9 points/year (SD 2.6, cut-points for highest and lowest quartiles 0 and 3.1 points/year) (Table 2), and 116 patients (44.4%) decreased less than 1 point/y (Fig. 2). Multiple imputation of missing MMSE values at follow-up resulted in a mean decline of 2.1 points/year for all patients and changed the proportion of slow progressors to 42.9%. The mean decline in IADL was –0.13/year (SD 0.14), and 163 patients (60.4%) showed little decline, of which 71 patients lost independence on only one IADL item from baseline to follow-up, 78 had unchanged score, and 14 improved.

OPEN IN VIEWER

Fig.2

Percentage of patients being slow progressors or experiencing intermediate or rapid progression, as measured by a global score (CDR-SB), a cognitive test (MMSE) and an assessment of function (IADL). CDR-SB, Clinical Dementia Rating scale sum of boxes; MMSE, Mini-Mental State Examination; IADL, Instrumental Activities of Daily Living (Lawton and Brody). Slow progression defined on CDR-SB and MMSE as <1 point worsened score per year, and on the IADL scale (Lawton and Brody) defined as loss of independence in maximally 1 IADL function over the follow-up period of mean 2 years. Intermediate or rapid progression defined as score differences from baseline to follow-up above the defined thresholds for progression.

Rapid progression, as defined by a decrease in MMSE score of 6 points or more per year, was found in 19 (7.3%) patients, whereas 116 (44.4%) were slow progressors, and 126 (48.3%) intermediate progressors. Better scores on cognitive tests and CDR-SB at baseline predicted slow progression as measured both with the MMSE and IADL (results not shown).

As shown in Table 5, the correlation between the change in CDR-SB and the change in IADL (–0.61) was somewhat stronger than that between the CDR-SB and MMSE (–0.53). The changes in both the cognitive and the functional subscore of CDR-SB correlated more strongly with IADL (–0.49 and –0.59, respectively) than with MMSE scores (–0.44 and –0.49, respectively). The weakest correlation was between changes in MMSE and IADL scores (0.36). When analyzed separately by CDR group, correlations were generally strongest for the CDR 1 group, for which correlations between change in CDR-SB and change in IADL and MMSE were –0.65 and –0.53, respectively. For the CDR 0.5 group correlations between change in CDR-SB and change in IADL and MMSE were –0.61 and –0.47, respectively. The CDR 2 group had the weakest correlation between change in CDR-SB and change in IADL (–0.43), while the correlation with change in MMSE was the strongest (–0.59).

DISCUSSION

This study investigates the progression of AD, predictors for progression and the correlation between change in a global score (CDR-SB), a cognitive (MMSE) and a functional measure (IADL). The overall progression rate of AD patients in the PADR study is similar to what has been reported in other studies. However, we found considerable heterogeneity, as half of the patients progressed slowly over the follow-up period of two years. Multiple regression analysis showed that the Trail Making Test B and the CERAD word-list delayed recall at baseline predicted progression, and could together with age, ApoE4 carrier status, hypertension, and the number of drugs used explain 17% of the variance in CDR-SB change. The correlation between changes in CDR-SB, MMSE and IADL scores seemed strongest between the CDR-SB and IADL and least strong between the MMSE and IADL scores.

Although AD is a progressive disorder, we found that a large proportion of patients experienced no or little decline during two years of follow-up, which is consistent with other studies [5, 9]. The large proportion of slow progressors illustrates the need to look beyond the mean progression rates to understand the considerable variance in disease course in AD. A fraction of AD patients show a rapid decline, a larger group experience an intermediate progression, but having slow disease progression may be equally, or even more, common. It has been argued that a high number of slow progressors in clinical AD studies may be due to selection of healthier participants in such studies [9]. However, a high proportion of slow progressors among AD patients was also found in a population study [6]. We believe our and similar studies deliver important background information both for patients in the symptomatic stages of the disease and for intervention studies in AD. The high proportion of slow progressors over the first two years implies a need for large study groups and/or long follow-up periods to ensure sufficient power to detect effects of interventions. Additionally, there is a need for more research to identify predictors of progression in AD and to explore whether there are biological differences that could explain the divergent disease courses.

Our study showed considerable heterogeneity in the disease course of AD patients. A similar study found that cognitive tests at baseline predicted 37% of the subsequent change in CDR-SB over three years of follow-up [26]. Although that study population was somewhat healthier, the results were similar [26], showing that lower scores on cognitive tests at baseline predicted an unfavorable disease course [9].

In the present study, the predictors included at baseline were able to explain only a small proportion (17%) of the variance in progression. We did not find any significant influence of sex, ApoE genotype, length of education, or medical history concerning other vascular risk factors or cerebrovascular disease on the disease course in AD. In the literature, a range of other factors have been examined for their potential influence on disease progression in the symptomatic phases of AD, including family history of dementia, neuropsychiatric symptoms, markers of inflammation, and functional level at baseline. The evidence for all these factors is mostly inconclusive, and even less is known about what influences progression in the preclinical phases of the disease [11, 12]. Further analyses will be performed in the PADR study focusing on other predictors for disease progression.

As in other studies, some of the study patients improved from baseline to follow-up [27], which could not be explained by depressive symptoms or the use of anti-dementia medication. It might be hypothesized that the improvement can be related to optimized treatment of comorbid disorders including withdrawal of inappropriate drugs, or it can be a consequence of test-retest variability in scores for patients who were essentially stable.

Most, but not all, studies in AD report an inverse association between the age at onset and rate of disease progression [7, 11]. In our study, older age was found to be a predictor of cognitive deterioration in the unadjusted regression analysis but could explain only a small fraction of the change in CDR-SB, and the effect of age disappeared in the adjusted analysis. Slow progressors in our study were on average younger. Studies of imaging and pathology in AD show faster evolving brain atrophy in early-onset than in late-onset AD [28], which is in contrast to our study where some of the younger patients with mild or very mild AD showed little progression. The heterogeneity of our sample regarding age and disease severity indicates that age-related comorbidities may be of importance for disease progression although this could not be shown by the analyses in the present study [29].

Although a majority of studies find that education may be a predictor of rapid cognitive decline in AD, others find no effect of education on progression [9, 11] which is in line with our findings. The mean education level in our study is relatively high compared to other clinical studies in AD. The younger patients with little progression also included many with long education. If well-educated patients and their families have an increased awareness of early AD symptoms, they may receive a diagnosis at an earlier stage, when disease progression is likely to be slow.

In contrast to some other studies, we did not find differences in CDR-SB progression rates across different stages of disease severity at baseline but were able to demonstrate such a difference for the MMSE, where the number of points lost increased with disease severity. There was also a difference in the number of slow progressors characterized by these scales, with the CDR-SB defining a higher number as slow progressors than the MMSE. This was also shown by Tschanz et al. [6] in another study using the same definitions of slow progression. The observed differences between MMSE and CDR-SB may have several explanations. While MMSE scores are known to depend on age and education [30], these factors are not known to affect the CDR-SB [31]. The MMSE is most sensitive to change in the middle to late stages of the disease [32], whereas the sensitivity of CDR-SB is not known to differ with disease severity and can also be scored for those who are unable to undergo cognitive tests. With our chosen cutoff, we defined more patients as slow progressors in IADL as compared to CDR-SB or MMSE scores. This is probably due to the psychometric qualities of the questionnaire, which may not have the necessary responsiveness to change [33].

The correlations shown between change in the CDR-SB and change in the MMSE and IADL are in agreement with two other studies with similar design [3, 4]. Notably, the correlations were stronger between CDR-SB and IADL than between CDR-SB and MMSE. This illustrates that follow-up assessments should focus both on cognition and function. Therefore we consider CDR to be the best of these tools to assess disease progression.

A strength of the present study is the relatively large sample size. In contrast to some studies assessing only patients able to attend the memory clinic, we also did follow-up assessments of patients with severe disease in nursing homes.

The major limitation of the study is that the generalizability is restricted by the inclusion criteria and the attrition rate. Memory clinic patients represent a selected population in terms of age, education level, and disease presentation, and for ethical reasons, we could not include patients with severe dementia at baseline. We have studied progression of AD among patients with MCI and mild to moderate dementia. Therefore, the results are not valid for prodromal or severe stages of the disease. There is still limited evidence on progression in preclinical AD. More research on progression in the prodromal phase is needed when designing pharmacological interventions among these patients.

It might be that we overestimate the number of slow progressors because patients with severe AD were not included, these would probably have a faster progression. Patients lost to follow-up scored lower on the Trail Making B test and may therefore have a more rapid progression. In addition, patients who died between baseline and follow-up were not included in the analyses because we did not know whether the cause of death was related to AD or to intercurrent disorders. Another limitation is the fact that we performed only one follow-up assessment, making non-linear estimates of progression impossible.

ETHICS APPROVAL AND CONSENT TO PARTICIPATE

The Regional Committee for Medical and Health Research Ethics for South East Norway approved the study (REC number 2011/531). All patients received oral and written information and gave written, informed consent to participate. Only patients with capacity to consent were recruited at baseline, as recommended by REC.