There were two major findings of this study: 1. Apo e4 was associated with smaller CA3&DG volumes in the whole control population and in the subgroup of old controls but not in the subgroup of young controls. Although total hippocampal volumes tended to be smaller in Apo e4 carriers compared to non-Apo e4 carriers, these differences were not significant. 2. AD with the Apo e4 allele had significantly smaller CA3&DG than AD without the Apo e4 allele. Furthermore, AD patients had smaller ERC, subiculum, CA1, CA1-2 transition volumes and consequently also total hippocampal volumes but not CA3&DG volumes than age-matched controls. Taken together, these results suggest that Apo e4 exerts a regional selective effect on CA3&DG in cognitively normal subjects and AD.
The first major finding was that Apo e4 showed a regional selective effect on CA3&DG in cognitively normal Apo e4 carriers who had smaller CA3&DG volumes than non-Apo e4 carriers. Previous neuroimaging studies assessing the influence of Apoe e 4 on hippocampal volume in cognitively healthy subjects in cross-sectional studies had inconsistent results. For example, Lind et al. (2006)
studied a non demented population of 60 subjects between 49–79 years and found significantly smaller right hippocampal volumes in Apo e4 carriers compared to non-Apo e4 carriers which were also associated with lower performance in hippocampal related memory tasks. Similarly, Den Heijer et al (2002)
found bilaterally reduced hippocampal volumes in 60 – 90 years old cognitively normal elderly Apo e4 carriers which persisted even after exclusion of subjects with evidence for mild memory impairment. In contrast, Jack et al. (1998)
studying 125 cognitively normal elderly controls (mean age 80 years), Reiman et al. (1998)
studying 33 middle aged controls (50 –60 years), Cohen et al (2001)
studying 25 elderly women with increased risk for AD either due to advanced age or a first degree relative with AD and finally Jak et al (2007)
studying 52 cognitively normal elderly subjects (63–92 years) all found that, although hippocampal volumes tended to be smaller in Apo e4 carriers compared to non-Apo e4 carriers, these differences did not reach statistical significance. There are several possible explanations for the inconsistent finding of an Apo e4 effect on total hippocampal volume in cognitively normal subjects in these neuroimaging studies, e.g. different definitions for cognitively normal or non-demented and thus different frequencies of subjects with preclinical AD in the study population, different age ranges of the study population and, since the effect of Apo e4 seems to be dose-dependent, different frequencies of Apo e4 homozygotes in the study population. In our study, the effect of Apo e4 seemed to be restricted to CA3&DG, none of the other subfields was significantly affected. Although CA3&DG represents a relatively large part of the total hippocampus and Apo e4 carriers had smaller total hippocampal volumes than non-Apo e4 carriers, this difference was not statistically significant. This suggests that subfield measurements might be more sensitive to detect subtle effects on the hippocampus, particularly if these effects are regionally selective.
The effect of Apo e4 on CA3&DG was not in all age groups equally present. Apo e4 carrier state was associated with a significant effect on CA3&DG in the total population and in the subgroup of older subjects but not in the subgroup of young Apo e4 carriers although they tended to have smaller CA3&DG than young non-Apo e4 carriers. A power analysis showed that given the number of young subjects and using the error estimates from the regression analysis we had only a 15% power to detect significant difference between young Apo e4 carriers and non-Apo e4 carriers at a significance level alpha = 0.05. This suggests that, although a subtle effect of Apo e4 in young subjects cannot be excluded, the effect of Apo e4 on CA3&DG probably accumulates over life time so that it only becomes manifest as a volume reduction detectable by in vivo imaging at an higher age. An alternative explanation is that Apo e4 renders CA3&DG more vulnerable towards non-Apo e4 related insults occurring later in life, e.g. increased oxidative stress or amyloid deposition. Interestingly, age had also a significant effect on CA3&DG in the total population and in the subgroup of older subjects but only if Apo e4 was included in the model (data not represented). The interaction between Apo e4 carrier state and age however was not significant, indicating, that age and Apo e4 carrier state act additively but independently on the CA3&DG volume. This is in contrast to the age effect on CA1 which was only observed in the whole population and was independent from the inclusion of Apo e4 carrier state in the model.
Our findings of Apo e4 effects on CA3&DG in older subjects are in good agreement with findings in animal and autopsy studies. Studies in transgenic mice for example have shown that while there is no difference in DG synaptic spine density between young Apo e3 and Apo e4 mice, synaptic spine density is significantly reduced in old Apo e4 mice compared to wild-type or Apo e3 mice (Cambon et al., 2000
, Ji et al., 2003
). A similar reduction of synaptic spine density in the DG region in Apo e4 carriers compared to Apo e3 carriers has also been found in an autopsy study of elderly cognitively normal humans (Ji et al., 2003
). Based on the similarities, it is tempting to speculate that the age related CA3&DG volume loss in our study reflects the age related loss of synaptic connectivity/neuroplasticity described in these studies. However it cannot be excluded that the volume loss in CA3&DG is caused by other negative effects associated with Apo e4 carrier state, e.g. reduced neurogenesis (Levi et al., 2005
), reduced protection against extrinsic or intrinsic neurotoxic insults, e.g. amyloid deposition (Horsburgh et al., 1999
, Buttini et al., 1999
, Levi and Michaelson, 2007
) or a combination thereof.
The second major finding was that AD with the Apo e4 allele had significantly smaller CA3&DG volumes than AD without the Apo e4 allele. Apo e4 enhances the negative effects of amyloid on neurons (Mahley et al. 2006
) and is associated with impaired neuronal repair processes (Weisgraber and Mahley, 1996
), therefore, we had assumed that its effect would be most prominent in regions with high intrinsic vulnerability to AD. Autopsy studies found CA1 to be the most severely affected hippocampal subfield in AD (West et al, 1994
, Roessler et al., 2002
; Price et al., 2001
; Fukutani et al., 1995
) and therefore we had expected to find smaller CA1 volumes in AD with Apo e4 than in AD without the allele, but that was not the case. One reason for this might be that we did not have enough statistical power to detect an effect on CA1 since the number of AD subjects in our study, particularly the number of non-Apo e4 AD subjects, was small. A power analysis based on the subjects in the analysis showed that we only had a power of 12% to detect CA1 volume differences of CA1between the Apo e4 carriers and non-carriers. This might indicate that either there is truly no effect of Apo e4 on CA1 or that our marking method is not sensitive enough to detect it. An alternative explanation for the absence of an effect of Apo e4 on CA1 could be that other effects overshadowed the Apo e4 effect. For example, it could be possible that the combined pathological effects of AD and aging (AD without Apo e4 were older than AD with Apo e4) on this subfield and consequently loss of neuropil and neurons are so severe at this advanced stage of the disease that additional negative effects due to Apo e4 genotype are obscured.
It is particularly interesting that it was CA3&DG, the only subfield which did not show an AD effect, which was smaller in AD with Apo e4. It has been demonstrated in autopsy studies of AD patients that the human hippocampus reacts with increased neurogenesis and formation of new, immature neurons in the DG and CA1 to the pathological processes in AD (Jin et al., 2004
). Since efficient neurogenesis requires an intact cholesterol metabolism, it seems plausible that those compensatory processes could be adversely affected by Apo e4 carrier state. However, such a general effect of Apo e4 would not explain the regional selectivity of the effect observed in our study. A recent study in transgenic mice offers an intriguing explanation for this regional preference. Levi and co-workers (2007)
found that conditions favoring neurogenesis are associated with a selective accumulation of intraneuronal Apo e in the DG in Apo e3 and Apo e4 mice. In the case of Apo e4 but not of Apo e3 mice, they also observed a regionally selective increase of intraneuronal soluble amyloid beta in the DG which was associated with signs of increased apoptosis and reduced neuronal density. A mechanism like this would be consistent with the fact that Apo e4 had a regional selective effect on CA3&DG in AD, although this is of course highly speculative. Additional studies in a larger population of AD subjects and if possible correlations with autopsy findings will be necessary to confirm the subfield specific findings of Apo e4 on CA3&DG respectively lack thereof on CA1 in this preliminary study..
This study has several limitations: 1. The sample size of the AD group, particularly the group of non-Apo e4 AD subjects was small. It will be necessary, to validate the findings regarding AD and Apo e4 in a larger study. 2. The study was cross-sectional in design and thus we cannot exclude cohort effects or that some controls were in the early stages of AD or another dementing disease associated with hippocampal atrophy. Furthermore, since the subjects were not particularly recruited for a study on Apo e4 effects, the Apo e4 and non Apo e4 groups were not optimally matched and thus we might have missed Apo e4 effects. 3. Hippocampal subfields were only marked in a relatively small region of the anterior hippocampus. Furthermore, except for the hypointense line, the boundaries between the subfields were based on arbitrarily defined rules and compromises were made to facilitate consistent marking (cf methods
section). Therefore, we cannot exclude that we missed Apo e4 effects with a regional preference for regions not included into the marked section, e.g. for the head or tail, or Apo e4 effects restricted to the ventral aspect of CA1 or subiculum/presubiculum. Nonetheless, the total volume of the hippocampal cross-section on which the hippocampal subfield were marked was highly correlated with the volume of the total hippocampus (Pearson correlation coefficient r = 0.76, p <0.0001). Therefore, we think that volume changes in this section are representative for hippocampal volume losses in diseases which are likely to affect the whole length of the hippocampus, as for example AD, aging or Apo e4.
In conclusion, these preliminary findings suggest that Apo e4 carrier state exerts a regionally selective effect on hippocampal subfields which is restricted to CA3&DG. This effect becomes more manifest with increasing age. In addition to this, we also found evidence of a regionally selective effect of Apo e4 on CA3&DG in AD. These effects are consistent with findings in animal and autopsy studies which describe a negative effect of Apo e4 allele in the dentate gyrus. The fact that aging, Apo e4 carrier state and AD show regionally selective effects on hippocampal subfields suggests that measurements of hippocampal subfields might be more sensitive to distinguish between different processes affecting the hippocampus than measurements of total hippocampal volume.