Predicting memory decline in normal elderly: Genetics, MRI, and cognitive reserve

Abstract

Major predictors of Alzheimer's disease (AD) include apolipoprotein E (APOE)-ɛ4, hippocampal atrophy on magnetic resonance imaging (MRI), and memory dysfunction prior to diagnosis. We examined 159 normal elderly subjects with MRI and the California Verbal Learning Test (CVLT); 84 returned for longitudinal follow-up 5 years later. Analyses at baseline revealed significant variance in hippocampal volume accounted for by cerebral volume and age but not by APOE isoform. However, interactions involving APOE isoform and laterality were observed. As hypothesized, an APOE × time interaction was revealed for CVLT long-delay free recall: APOE-ɛ3/4 subjects had significantly poorer performance than APOE-ɛ3/3 subjects at follow-up. Forward stepwise multiple regression analysis predicting follow-up long-delay free recall selected baseline recall, followed by number of APOE-ɛ4 alleles, followed by left-hippocampal volume. Age and sex did not enter into the model. We conclude that APOE-ɛ4 predicts longitudinal memory decline in healthy controls and that MRI morphometry of hippocampus adds slightly to predictive value.

Introduction

Major predictors of Alzheimer's disease (AD) include apolipoprotein E (APOE)-ɛ4 [8], [9], [10], [39], [40], [52], hippocampal atrophy [1], [16], [33], [41], [42], [43], [50], [51], [59], [76], [102], and memory dysfunction prior to diagnosis [2], [8], [39], [47], [52], [59]. Considerable work has therefore been devoted to elucidating APOE-ɛ4 effects on hippocampal volume and memory performance to facilitate early detection and accuracy of AD diagnosis.

Cross-sectional magnetic resonance imaging (MRI) studies of individuals with APOE-ɛ4 compared with non-ɛ4 individuals have reported a smaller right [54] and bilateral [3], [30], [35] hippocampus in AD patients, reduced left-hippocampal volume in subjects with age-associated memory impairment (AAMI) [87], left-greater-than-right hippocampus in a mixed AAMI and normal sample [90], and smaller right [96], left [87], and bilateral [21], [55], [73] hippocampus in healthy controls. Others, however, have failed to find cross-sectional hippocampal differences between APOE-ɛ4 and non-ɛ4 controls [63], [77], [84], and Bigler et al. [5] did not find volumetric asymmetry in their study including a range of cognitive and neuropsychiatric impairments, leading them to question APOE-ɛ4 effects on hippocampus, particularly lateralized reductions [29]. Some studies likewise have reported that AD patients with APOE-ɛ4 exhibited poorer performance on memory tasks [36], [54], [66], whereas others have not [4], [52], [53]. Many (reviewed by Nilsson et al. [67]) but not all [4], [7], [14], [15], [25], [45], [61], [67], [72], [73], [80] cross-sectional studies of non-demented middle-aged and elderly subjects with APOE-ɛ4 have reported poorer learning and memory [8], [9], [13], [21], [27], [44], [56], [78], [84], [92] and other impairments in cognitive performance [6], [32], [44], [82], [101].

Longitudinal MRI studies of hippocampal changes in association with APOE-ɛ4 have been relatively sparse but have suggested that APOE-ɛ4 is associated with accelerated hippocampal-volume loss in healthy controls [15], [63]. Most neuropsychological studies have shown that APOE-ɛ4 is associated with declining memory [2], [6], [23], [36], [38], [44], [61], [88] and other cognitive processes [6], [20], [23], [44], [88], but null findings have also been reported [11], [15], [52], [72]. Some studies have reported no cross-sectional differences in memory but significant longitudinal differences [6], [61], [67], [79], [89]; others have reported the reverse [101]. One study reported a decline in cognition for women only [64]. Dik et al. [22] reported that APOE-ɛ4 was associated with memory decline in cognitively impaired subjects but not in normal subjects after a 3-year interval. Findings relating APOE-ɛ4 to memory performance in healthy individuals are therefore equivocal.

Although a number of studies have separately examined APOE-ɛ4 in relation to hippocampal volume or to memory, few studies have evaluated the predictive value of all three in combination. Marquis et al. [59] found hippocampal volume and memory performance to predict the development of questionable dementia, but APOE-ɛ4 did not add to the predictive value. The relative contribution of hippocampal volume versus memory performance to diagnosis is also uncertain, with some investigators questioning the value of hippocampal morphometry [50], [85]. The use of APOE genotyping alone in the diagnosis of AD is not supported [60], [94], [97], and thus combination studies are warranted examining the relative value of APOE isoform status and hippocampal morphometry in predicting memory decline.

We therefore studied the relationship between APOE-ɛ4, hippocampal volume, and both cross-sectional and longitudinal memory performance over a 5-year interval. Because delayed recall has been shown to be the most sensitive memory-related predictor of the development of AD [8], [47], [48], [56], [100], we hypothesized that possession of APOE-ɛ4 would be related to a longitudinal decline in California Verbal Learning Test (CVLT) long-delay free recall [8], [9]. We also hypothesized that hippocampal volume at baseline would predict CVLT performance at follow-up and that the combination of APOE-ɛ4 and hippocampal volume would be superior in predicting follow-up memory scores relative to APOE-ɛ4 alone.