Results demonstrate reduced volume of anterior MTL structures in presymptomatic mild cognitive impairment (pMCI), approximately 4 years prior to MCI diagnosis. Previous studies have reported volumetric reduction of the ERC and hippocampus in MCI compared to normals and in AD compared to MCI (Jack et al., 1997
; Convit et al., 1997
; Killiany et al., 2002
; de Leon et al., 2004
). Some studies have indicated more widespread volumetric reduction in MCI and mild AD, affecting neocortical regions (Chetelat et al., 2002
; Gold et al., 2005
; Pennanen et al., 2005
) and whole brain volume (Jack et al., 2004
; Fotenos et al., 2005
), suggesting tissue loss may already be occurring in a widespread manner throughout the brain in MCI.
In contrast, the present pMCI group had relatively small WBV reduction (2%) but considerable reduction in some MTL structures (19% in the hippocampus and 16% in the ERC). The pMCI reductions observed in the hippocampus and ERC are smaller than the range of reductions of approximately 30–50% reported in mild to moderate AD compared to normal groups (Killiany et al., 2002
; Pennanen et al., 2004
), consistent with reports that these structures undergo accelerated reduction in AD (Jack et al., 2000
; Pennanen et al., 2004
). In a study of 5 middle-aged adults who developed familial AD within 3 years, an average of 16% volume decrease in MTL structures was observed compared to controls (Schott et al., 2003
), which is comparable to the reductions we observed in seniors who developed the prodromal stage (MCI) of late-onset AD within 4 years.
The present results concur with several recent findings of reduced MTL volume in pMCI (den Heijer et al., 2006
; Smith et al., 2007
). Our results concerning the extent of hippocampal reduction are similar to those of den Heijer et al. (2006)
, who observed an average of 17% hippocampal reduction in seniors destined to develop dementia (AD and vascular dementia types) within 2–3 years (den Heijer et al., 2006
). However, this study did not compute volumes of hippocampus subregions, ERC, or WBV. By computing the volume of these structures, and parcellating the hippocampus into anterior (hippocampus head; HH), middle (hippocampus body; HB) and posterior (hippocampus tail; HT) regions, the present study demonstrated a gradient of MTL volumetric reduction in pMCI. In addition to the substantial ERC reductions noted above, the pMCI group had considerable reductions in the HH (25%) and HB (16%), but minimal reduction in the HT (1%) or WBV (2%). The exact reductions observed in hippocampal subregions should not be taken literally because they are likely to vary somewhat depending upon the specific boundaries used for segmentation. Nevertheless, the general finding of an anterior-to-posterior gradient of hippocampal reductions is in-line with pathology data showing that AD begins in anterior portions of the MTL before affecting posterior MTL and neocortex (Braak and Braak, 1995
Results from the present study also demonstrate that normalized anterior MTL volume can differentiate individuals destined to develop MCI within a few years from those who will remain cognitive normal with a high degree of accuracy. After controlling for age, sex, education level, and ICA, volumes of anterior portions of the hippocampus (HH and HB) differentiated the pMCI and normal groups with 90% accuracy. The volumes of these structures were more accurate in differentiating the pMCI and normal groups than memory tests known to be sensitive to amnestic MCI (the WMS and CERAD word list delayed), which were 80% accurate. However, the highest discriminatory accuracy (93%) was observed when these memory scores were included in a model with anterior hippocampal volumes.
Although pMCI participants had lower average WMS and CERAD word list delayed scores than the normal group at baseline, there are several reasons for which they were unlikely to have had undetected MCI. First, the baseline cognitive scores of all pMCI participants were within the age- and education-adjusted normal range and none of the participants had either diagnostic criteria necessary for MCI diagnosis: a documented memory complaint, and a CDR box score of 0.5 for memory. Second, it is not uncommon for normal individuals who later develop AD to have lower average cognitive test scores than their age-matched peers who remain normal, sometimes extending back to childhood (Linn et al., 1995
; Elias et al., 2000
; Whalley et al., 2000
). Finally, the average MMSE score for the pMCI group at MCI diagnosis was 27.7, a near-normal average score approximately 4 years after their baseline scan.
Noninvasive in vivo probes are needed to identify seniors at risk for cognitive decline because emerging interventions may be most successful prior to MCI (Smith, 2007
). Baseline measures are likely to contribute to this process due to the time-sensitive need to establish diagnosis. The present results suggest that baseline volumetric measurement can help identify individuals at future risk for MCI. In particular, results suggest that anterior MTL structures may be of particular diagnostic relevance because reduction of these structures was observable in pMCI participants compared to normal participants who had a longer average follow-up time from scan to diagnosis. The observation that volumetric declines were relatively confined to anterior MTL structures in pMCI offers hope that potential therapies could protect the brain from reductions of posterior MTL and WBV associated with the onset of MCI.
This study had several limitations that highlight open questions and may help guide future research. First, like most other normal control volunteer groups at AD centers, participants were highly educated. High education may buffer effects of brain pathology on cognition (Stern, 2002
), which could lengthen the presymptomatic period relative to groups with lower education. Future studies will be required to determine the extent to which the present and similar findings generalize to cohorts with lower levels of education. Our ADC is currently in the process of attempting to recruit lower education volunteers in the normal control cohort to address this question. Second, the cross-sectional nature of the study does not enable us to determine the relationship between cognitive decline and volumetric reductions in various ROIs over time. We are in the process of following pMCI participants longitudinally to address this issue. Third, as with most studies of this kind, the sample size of the presymptomatic group was relatively small and pathological diagnosis cannot yet be confirmed. Although most individuals given a clinical diagnosis of MCI at our center show AD pathology (Markesbery et al., 2006
), it is not yet possible to know what percentage of the present MCI or normal samples harbor significant AD pathology. Some misclassification of participants (according to the neuropathology gold standard) is thus possible. This issue should be addressable in the future because participants in the longitudinal normal cohort study at our ADRC have agreed to brain donation at death.