Introduction
Down's syndrome (DS) is associated with trisomy of chromosome 21, and is the most common genetic cause of learning disability (mental retardation), occurring in approximately 1 in 1000 live births (Roizen & Patterson, Reference Roizen and Patterson2003). Individuals with DS are prone to developing dementia (most commonly Alzheimer's disease; AD) in middle age. The combination of pre-existing learning disability with superimposed dementia in people with DS is difficult to assess and treat, and is an expensive management problem.
The neuropathology of AD and some degree of the gross neuroanatomical changes associated with AD have been reported to occur in the brains of almost all DS people older than 40 years (Malamud, Reference Malamud and Gaitz1972; Wisniewski et al. Reference Wisniewski, Wisniewski and Wen1985). In addition, the prevalence of clinically detectable dementia in people with DS is reported to rise to 66% in the sixth decade of life (Visser et al. Reference Visser, Aldenkamp, van Huffelen, Kuilman, Overweg and van Wijk1997). The amyloid precursor protein (APP) gene is localized to chromosome 21. Thus it has been hypothesized that the presence of an extra copy of the APP gene in DS cells leads to abnormalities in APP processing in neuronal membranes, and then to the formation of amyloid plaques, neuronal death and clinical AD (Prasher et al. Reference Prasher, Farrer, Kessling, Fisher, West, Barber and Butler1998; Folin et al. Reference Folin, Baiguera, Conconi, Pati, Grandi, Parnigotto and Nussdorfer2003).
The presence of pre-morbid learning disability in people with DS may make additional cognitive deficits associated with the early stages of AD more difficult to detect than in the general population (Miniszek, Reference Miniszek1983). Treatment for AD is more successful if it is begun at an early stage; however, there are few non-invasive aids to diagnosis. Magnetic resonance imaging (MRI) may provide a technique whereby differences in brain anatomy can be detected and used to aid the early diagnosis of AD in DS.
Four volumetric MRI studies have compared whole brain anatomy in DS individuals with and without dementia (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998; Aylward et al. Reference Aylward, Li, Honeycutt, Warren, Pulsifer, Barta, Chan, Smith, Jerram and Pearlson1999; Prasher et al. Reference Prasher, Cumella, Natarajan, Rolfe, Shah and Haque2003). These studies reported that DS individuals with AD, as compared to non-demented DS controls, have a significant reduction in volume of medial temporal lobe/hippocampus (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998; Aylward et al. Reference Aylward, Li, Honeycutt, Warren, Pulsifer, Barta, Chan, Smith, Jerram and Pearlson1999) and neocortex (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994), together with significant enlargement of ventricular cerebrospinal fluid (CSF) (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998; Prasher et al. Reference Prasher, Cumella, Natarajan, Rolfe, Shah and Haque2003). These reports are broadly consistent with the pattern of gross brain deficits found in AD in the general population.
These prior MRI studies of DS and dementia were important first steps. However, they examined relatively small numbers of demented DS individuals (ranging from 2 to 11) and most used demented DS participants who were significantly older than the non-demented DS controls (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998; Aylward et al. Reference Aylward, Li, Honeycutt, Warren, Pulsifer, Barta, Chan, Smith, Jerram and Pearlson1999). Recruiting age-matched samples of DS individuals with and without AD for a MRI study is difficult, as the population of older DS individuals is small, and this is a group of people who often find the demands of MRI scanning difficult. However, age is a potentially important confound because non-demented DS individuals are reported to have significant age-related decreases in regional brain volumes, including medial temporal volumes (Aylward et al. Reference Aylward, Li, Habbak, Warren, Pulsifer, Barta, Jerram and Pearlson1997; Krasuski et al. Reference Krasuski, Alexander, Horwitz, Rapoport and Schapiro2002; Teipel et al. Reference Teipel, Schapiro, Alexander, Krasuski, Horwitz, Hoehne, Möller, Rapoport and Hampel2003, Reference Teipel, Alexander, Schapiro, Möller, Rapoport and Hampel2004). Despite this, two of the prior MRI studies of demented DS individuals did not correct for age in their statistical analysis (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998) and thus their results may not be reliable.
To date, no study has investigated whether overall brain anatomy is related to cognitive function in people with DS and AD. Thus, the neurobiological associates of cognitive deficits in DS individuals with dementia/AD are poorly understood. We therefore used volumetric MRI to compare whole brain anatomy in DS individuals with and without AD (correcting for head size, age and gender). We also determined whether brain volumes could reliably classify individuals according to AD status. Furthermore, within DS individuals with AD, we investigated whether brain anatomy was significantly related to cognitive function, as measured by the Cambridge Cognitive Examination (CAMCOG; Huppert et al. Reference Huppert, Brayne, Gill, Paykel and Beardsall1995; Roth et al. Reference Roth, Huppert, Mountjoy and Tym1998).
Method
Participants
We studied 19 DS adults with AD, and 39 non-demented, healthy adults with DS. Participants with DS were recruited locally, and from already identified cohorts in Birmingham, Plymouth and Newcastle, UK. DS status was assessed in all participants by karyotyping. Dementia status was assessed using the Diagnostic Criteria for Research-10 (DCR-10; WHO, 1992). As expected, the mean age of the AD group was significantly higher than the healthy DS group: 52 years (range 42–62 years) and 35 years (range 19–66 years) respectively. Nineteen of the 38 healthy DS participants were older than 35 years at the time of scanning. There was also a higher proportion of female participants in the AD group (10 of 19 in the AD group, as compared to 12 of 39 in the healthy DS group). However, the two groups were similar in terms of ethnic origin, apolipoprotein E (apoE) status and handedness (see Table 1).
Group comparisons for Down's syndrome (DS) adults with Alzheimer's disease (AD), and healthy DS controls. p values are stated after correcting for total cranial volume, age and gender (standard deviations are given in parentheses)
CAMCOG, Cambridge Cognitive Examination; WBV, whole brain volume; CSF, cerebrospinal fluid; n.a., not applicable.
* Significant at p<0.05.
All participants underwent standard physical, neurological and psychiatric screening, including routine bloods (e.g. renal and liver function tests, full blood count and thyroid function tests) and clinical MRI. Psychiatric screening was performed by a psychiatrist (V.P.) and a psychologist (F.B.). We excluded people with untreated physical disorder affecting brain function, known history of birth trauma or head injury and major psychiatric condition other than AD. In addition, we excluded participants whose clinical MRI scans suggested acquired brain damage.
None of the participants was taking antipsychotic or antidepressant medication at the time of the study; however, nine of the 19 DS participants with AD were taking cholinesterase inhibitors. Some participants were also taking other medications for unrelated physical conditions (e.g. hypertension), but this did not differ significantly between groups.
Multi-Centre Research Ethics Committee (MREC) and Local Research Ethics Committee (LREC) approval was obtained, and after complete description of the study to the participants, written informed consent, or assent, was obtained from the participants and/or their carers.
MRI protocol
All participants were scanned using a 1.5 T GE Signa MR NV/i system (General Electric, Milwaukee, WI, USA) at the Maudsley Hospital, London, and were interleaved in the dates of their scans. A vacuum fixation device was used to ensure that participants were both comfortable and restrained from movement during the scanning process. The whole brain was scanned with a three-dimensional (3-D) inversion recovery prepared fast spoiled gradient-recalled (SPGR) acquisition in the steady-state longitudinal relaxation time (T1)-weighted dataset. These T1-weighted images were obtained in the axial plane with 1.5-mm contiguous sections. Repetition time (TR) was 13.8 ms, inversion time was 450 ms, echo time (TE) was 2.8 ms, and the flip angle was 20° with one data average and a 256×256×124 pixel matrix. Image contrast for all datasets was chosen with the aid of a software tool for optimizing image contrast (Simmons et al. Reference Simmons, Arridge, Barker and Williams1996). Acquisition time was 6 min 27 s.
Volumetric protocol
Volumetric analysis of total and regional brain areas was performed on a reformatted SPGR dataset using Measure software (Barta et al. Reference Barta, Dhingra, Royall and Schwartz1997). Right and left hemispheric brain matter, total cranial volume and cerebral ventricles were measured on images aligned along the anterior/posterior commissure line. Measurements were then made, using previously described region of interest boundaries (Murphy et al. Reference Murphy, DeCarli, McIntosh, Daly, Mentis, Pietrini, Szczepanik, Schapiro, Grady, Horwitz and Rapoport1996), of total, right and left total cranial volume, whole brain, frontal, prefrontal, temporal, parietal and occipital lobes, hippocampus, amygdala, cerebellum, caudate, putamen, striatum (caudate+putamen), lateral ventricles, peripheral CSF and total ventricular space (lateral+third ventricles). Images were realigned parallel to the Sylvian fissure for hippocampal and amygdalar measures.
The volume of each region was calculated by multiplying the summed pixel cross-sectional areas by slice thickness. All volumetric measurements were made by a single rater who was blind to the status of each participant. Intra-class reliability was determined for all brain volumes: these were all highly significant (total cranial volume: r=0.963; whole brain volumes: r=0.977; hippocampus: r=0.911; amygdala: r=0.902; caudate: r=0.923; putamen: r=0.916; striatum: r=0.909; frontal lobes: r=0.956; prefrontal: r=0.949; temporal lobes: r=0.972; parietal lobes: r=0.917; occipital lobes: r=0.943; cerebellum: r=0.969; lateral ventricles: r=0.942; peripheral CSF: r=0.931; total ventricular space: r=0.975; all p values <0.001).
Cognitive assessment
Cognitive ability was measured using the CAMCOG (Huppert et al. Reference Huppert, Brayne, Gill, Paykel and Beardsall1995; Roth et al. Reference Roth, Huppert, Mountjoy and Tym1998). The CAMCOG has been validated for use with DS adults (Hon et al. Reference Hon, Huppert, Holland and Watson1999) and provides a measure of general cognitive function, including measures of memory, orientation, language, attention, praxis and executive function. The CAMCOG is appropriate for assessing cognitive function in people with learning disability, unlike more standard tests of cognitive function such as the Wechsler Adult Intelligence Scales. CAMCOG subtests, with a small number of exceptions, did not produce ceiling or floor effects. The CAMCOG incorporates, and is highly correlated with, the Mini-Mental State Examination (MMSE; Blessed et al. Reference Blessed, Black, Butler and Kay1991). For each participant cognitive testing was conducted within 6 months of scanning.
Statistical analysis
Analysis of brain volumes was carried out using SPSS 8.0 for Windows (SPSS Inc., Chicago, IL, USA). Normality of distribution was assessed in both groups, and tested for significance using the Kolmogrov–Smirnov statistic. Neither group violated the assumption of normality, therefore parametric tests of difference and correlation were used.
Group differences in brain volumes were tested with one-way analysis of co-variance (ANCOVA), with group as the between-participant factor and total cranial volume, participant age and gender as covariates.
We examined the relevant scatterplots and there was no evidence of non-linearity. Therefore, Pearson's correlation coefficient (two-tailed) was used to assess linear correlations between regional brain volumes and age, in both groups.
Within DS individuals with AD, we also carried out an exploratory post hoc analysis (using Pearson's test of correlation) of the relationship between CAMCOG scores and those brain volumes that differed significantly between groups (when corrected for total cranial volume, age and gender).
Our results may have been confounded by medication status, therefore we compared brain anatomy and cognitive function between those demented individuals taking cholinesterase inhibitors to those who were not.
We used logistic regression to assess whether brain volumes could reliably classify individuals according to AD status.
The level of statistical significance was defined as p<0.05 (two-tailed).
Results
Most uncorrected brain volumes (all except occipital lobe and cerebellum) were significantly smaller in the DS+AD group, as compared to the healthy DS group (see Table 1). However, as noted above, brain volumes may be confounded by differences in head size, age and gender.
Mean total cranial volume was significantly larger in the non-demented group (for total cranial volume: F=8.8, df=1, 57, p=0.004).
As noted earlier, there was a larger proportion of females in the DS+AD group. Within the healthy DS group there were some significant effects of gender on brain volumes: males had significantly larger volumes of left whole brain (t=2.23, p=0.032), left prefrontal lobes (t=2.27, p=0.029), left and right temporal lobes (t=4.14, p<0.001; t=3.12, p<0.001) and right caudate (t=2.24, p=0.032). Within the DS+AD group there were no significant effects of gender on brain volumes or level of cognitive function. In addition, within the DS+AD group there were no significant effects of medication on brain volumes or level of cognitive function.
After correcting for group differences in total cranial volume, age and gender, DS individuals with AD compared to healthy DS individuals had a significantly smaller volume of left and right hippocampus, right amygdala, left and right caudate, right putamen, left and right putamen, and left peripheral CSF (see Table 1).
As expected, DS individuals with AD had significantly lower scores on most cognitive measures, compared to healthy DS individuals, both before and after correcting for age. These included CAMCOG total score (F=8.72, df=1, 57, p=0.005, after correcting for age) and CAMCOG total memory score (F=10.41, df=1, 57, p=0.02, after correcting for age).
Within DS individuals with AD, there were no significant correlations between those regional brain volumes that differed significantly between groups and measures of cognitive function (when corrected for total cranial volume, age and gender).
Logistic regression analysis showed that hippocampal volume (and using total cranial volume, age and gender as covariates) was able to correctly categorize 92% of the non-demented DS individuals (specificity) and 75% of DS individuals with AD (sensitivity) (see Fig. 1). Similar analysis showed that caudate volume (again using total cranial volume and age as covariates) were able to correctly categorize 92% of the non-demented DS individuals (specificity) and 80% of DS individuals with AD (sensitivity) (see Fig. 2).
Mean hippocampal volumes in healthy individuals with Down's syndrome (DS) and individuals with DS and Alzheimer's disease (AD).
Mean striatal volumes in healthy individuals with Down's syndrome (DS) and individuals with DS and Alzheimer's disease (AD).
Discussion
In this observational study we found that DS individuals with AD had significantly smaller uncorrected volumes of most brain regions we measured, as compared to non-demented DS participants. However, these initial findings were potentially confounded by significant between-group differences in head size, age and gender. To overcome the potential confound of head size, we included total cranial volume as a covariate in our statistical analysis. We also used age as a statistical covariate. This was necessary because some prior volumetric MRI studies of brain ageing in healthy DS individuals have reported significant age-related decreases in volume of a number of brain regions, including hippocampus (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Krasuski et al. Reference Krasuski, Alexander, Horwitz, Rapoport and Schapiro2002), amygdala (Krasuski et al. Reference Krasuski, Alexander, Horwitz, Rapoport and Schapiro2002), caudate nucleus (Raz et al. Reference Raz, Torres, Briggs, Spencer, Thornton, Loken, Gunning, McQuain, Driesen and Acker1995) and putamen (Aylward et al. Reference Aylward, Li, Habbak, Warren, Pulsifer, Barta, Jerram and Pearlson1997). Finally, we included gender as a covariate. This was necessary, first, because the DS+AD group was composed of a greater proportion of females compared to the healthy DS group. Second, there were a small number of sex-related differences in brain volumes within the healthy DS group, although none were present within the DS+AD group.
After correcting for the potential confounds of head size, age and gender, we found that the DS+AD group had a significantly smaller volume of hippocampus, striatal structures (caudate and putamen) and amygdala. Furthermore, we report for the first time that volume of hippocampus and caudate nuclei can distinguish between DS people with and without AD, with good specificity (90%) and reasonable sensitivity (80%).
Another potential confound for our results is that some DS individuals with AD were taking cholinesterase inhibitors and some were not. The reasons for this discrepancy are unknown, but it most probably reflects differences in local prescribing habits. We cannot state whether or not medication status in our DS+AD sample was related to length of illness. This is because it is often difficult to accurately assess the date of onset of dementia in people with DS, and we were not able to retrospectively establish the time of AD onset or the length of illness.
We found no significant differences in any clinical variable between those demented DS participants who were taking cholinesterase inhibitors and those who were not. Nonetheless, it remains possible that medication status was a significant confound for our analysis. However, many people with DS and AD are treated with cholinesterase inhibitors, and therefore our findings may be more generalizable to the whole DS population than if we had excluded these participants.
We carried out multiple statistical tests, and so some of our results may be confounded by type 1 error. Nevertheless, our finding of group differences in regional DS brain anatomy in DS participants with AD is in broad agreement with prior smaller reports. In addition, most of the significant group differences we found survived Bonferroni correction for multiple comparisons (left hippocampus, right caudate, left and right striatum, left peripheral CSF). The Bonferroni correction is a stringent correction for this dataset because volumes of the different brain regions tend to be positively correlated. Thus, it is possible, but unlikely, that our findings can be explained by type 1 error.
We used clinical rather than post-mortem criteria for assessing AD. Therefore, we cannot exclude the possibility that some DS participants in the AD group had mixed neuropathology. Nonetheless, the DS individuals with AD that we included were well characterized, and we excluded people with easily detectable physical health problems. Therefore, the group differences we found in brain anatomy most probably reflect the development of AD-type neuropathology, rather than other types of neuropathology.
We used manual delineation methods to measure brain volumes, rather than voxel-based morphometry (VBM), a fully automated technique for examining whole brain morphology. VBM uses statistical modelling assumptions that may not apply to populations with marked abnormalities in brain anatomy. Therefore, VBM analysis of our data may not have been reliable.
As noted above, nearly all older people with DS have the neuropathology of AD, but not all older DS individuals display clinical symptoms of the disorder. Hence, it is very likely that the brains of many of our clinically non-demented DS controls were significantly affected by plaques and tangles. Therefore, the differences we report in brain anatomy cannot be taken to specifically reflect the underlying neuropathology of AD. Rather, these differences in brain anatomy are related to clinical AD status.
Hippocampus and amygdala
Our finding that DS individuals with AD, as compared to healthy DS individuals, had a significantly smaller corrected hippocampal volume is consistent with a report that the hippocampus is one of the brain regions most severely affected by amyloid plaques and non-functioning tumours (NFTs) in the DS brain (Hof et al. Reference Hof, Bouras, Perl, Sparks, Mehta and Morrison1995). It is also consistent with two previous volumetric MRI studies of DS individuals with dementia (Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998; Aylward et al. Reference Aylward, Li, Honeycutt, Warren, Pulsifer, Barta, Chan, Smith, Jerram and Pearlson1999) and a number of MRI studies of AD patients in the general population (e.g. Double et al. Reference Double, Halliday, Kril, Harasty, Cullen, Brooks, Creasey and Broe1996; Karas et al. Reference Karas, Burton, Rombouts, van Schijndel, O'Brien, Scheltens, McKeith, Williams, Ballard and Barkhof2003; Pennanen et al. Reference Pennanen, Kivipelto, Tuomainen, Hartikainen, Hänninen, Laakso, Hallikainen, Vanhanen, Nissinen, Helkala, Vainio, Vanninen, Partanen and Soininen2004).
Our results suggest that reduction in hippocampal volume may provide a useful tool for the diagnosis of AD in people with DS, as has been proposed for AD in the general population (e.g. Laakso et al. Reference Laakso, Partanen, Riekkinen, Lehtovirta, Helkala, Hallikainen, Hanninen, Vainio and Soininen1996; Yamaguchi et al. Reference Yamaguchi, Meguro, Shimada, Ishizaki, Yamadori and Sekita2002). For a brain region to be a useful indicator of a disease, it must be relatively unaffected by normal ageing but substantially affected in the diseased state. One previous study reported that hippocampal volume is relatively stable in healthy, non-demented DS adults (Aylward et al. Reference Aylward, Li, Honeycutt, Warren, Pulsifer, Barta, Chan, Smith, Jerram and Pearlson1999), as has also been reported in the general population (e.g. Sullivan et al. Reference Sullivan, Marsh, Mathalon, Lim and Pfefferbaum1995). However, other volumetric MRI studies of non-demented individuals with DS have reported significant age-related reductions in hippocampal volumes (Kesslak et al. Reference Kesslak, Nagata, Lott and Nalcioglu1994; Krasuski et al. Reference Krasuski, Alexander, Horwitz, Rapoport and Schapiro2002). In our own sample of non-demented DS individuals there were no significant age-related differences in bulk-volume of the hippocampus. Accordingly, in our dataset hippocampal volume was effective in being able to discriminate between DS individuals with and without AD. Corrected total hippocampal volume was able to correctly categorize 92% of the non-demented DS individuals and 75% of DS individuals with AD. Thus, our results provide preliminary evidence that hippocampal volume may be of diagnostic value for individuals with DS, in particular to exclude the diagnosis of AD. We do not claim that volumetric MRI can be used as a stand-alone diagnostic tool for dementia in DS people. Rather, we suggest that in complex cases measurement of specific brain regions using MRI, when combined with careful clinical examination and recently published scales (Deb et al. Reference Deb, Hare, Prior and Bhaumik2007), will allow more accurate diagnosis and therefore more appropriate treatment.
We also found that DS individuals with AD had a significantly smaller corrected volume of right (but not left) amygdala, as compared to non-demented DS participants. The significance of this difference being only on the left side is unknown. However, this finding is consistent with one previous study that also reported a significant reduction in left, but not right, amygdala volume in demented DS individuals compared to non-demented DS controls (Pearlson et al. Reference Pearlson, Breiter, Aylward, Warren, Grygorcewicz, Frangou, Barta and Pulsifer1998). More generally, our finding of a reduction in amygdala volume is consistent with volumetric studies of AD in the general population (e.g. Busatto et al. Reference Busatto, Garrido, Almeida, Castro, Camargo, Cid, Buchpiguel, Furuie and Bottino2003).
Caudate and putamen
DS individuals with AD had a significantly smaller volume of caudate, putamen and total striatum (caudate+putamen), as compared to healthy DS controls. Reductions in striatal volumes have not been reported previously in DS individuals with AD. However, significant volume reductions in caudate (but not putamen) have been reported by a number of VBM studies of AD in the general population (Rombouts et al. Reference Rombouts, Barkhof, Witter and Scheltens2000; Frisoni et al. Reference Frisoni, Testa, Zorzan, Sabattoli, Beltramello, Soininen and Laakso2002; Karas et al. Reference Karas, Burton, Rombouts, van Schijndel, O'Brien, Scheltens, McKeith, Williams, Ballard and Barkhof2003).
The caudate and putamen are both classically thought to be involved in organizing and guiding complex motor function (e.g. Marsden, Reference Marsden1982). However, the striatum receives inputs from all cortical regions, and is also thought to be involved in learning and memory (Poldrack et al. Reference Poldrack, Prabhakaran, Seger and Gabrieli1999; Packard & Knowlton, Reference Packard and Knowlton2002). It is possible, therefore, that reductions in striatal volumes may contribute to memory and/or other cognitive deficits in DS people with AD.
Within our sample of non-demented DS individuals, putamen volume was subject to a significant age-related decrease, but caudate volume was stable with respect to age. This suggests that reductions in caudate volume are more likely than reductions in putamen volume to provide a diagnostic marker for AD in people with DS. In addition, in our sample corrected total caudate volume was able to correctly categorize 92% of the non-demented DS individuals and 80% of DS individuals with AD. Thus, volume of caudate (but not putamen) may be of diagnostic value to exclude the diagnosis of AD.
CSF
DS individuals with AD had a significantly larger corrected volume of left (but not right) peripheral CSF, as compared to non-demented DS controls. Peripheral CSF is a marker of cortical atrophy. This may suggest that, for unknown reasons, cortical atrophy in people with DS is greater in the left hemisphere than in the right.
We did not find that DS individuals with AD had significant differences in corrected volume of lateral ventricles or total ventricular space (lateral+third ventricles), as compared to non-demented DS controls. Thus, we did not replicate the results of a previous study that reported that DS individuals with AD had a significant increase in total ventricular volume, as compared to age-matched, non-demented DS controls (Prasher et al. Reference Prasher, Cumella, Natarajan, Rolfe, Shah and Haque2003). Ventricular enlargement is a cardinal feature of AD in the general population. Our failure to detect significantly greater total ventricular volume in DS individuals with AD may indicate that the clinical presentation of AD in DS individuals requires a proportionally smaller reduction in total brain volume than in AD in the general population. Thus, the expression of the clinical phenotype of AD in DS may depend to a greater degree on medial temporal volume loss, which we did detect. In addition, in the general population the clinical presentation of AD may require additional loss of whole brain volume, as non-learning disabled individuals have a greater cognitive ‘reserve’. Another possibility is that our DS+AD sample was composed of individuals with the equivalent of relatively mild AD, as more severely demented individuals with DS are generally unable to comply with the demands of MRI scanning. The differences in regional brain anatomy we found may therefore correspond only to mild AD in the general population.
Relationship between brain anatomy and cognitive function
We found that DS individuals with AD had significant reductions in volume of medial temporal regions (hippocampus and amygdala) and striatal structures (putamen and caudate). In healthy populations the hippocampi and amygdalae are implicated in memory function (Tranel & Hyman, Reference Tranel and Hyman1990; Squire et al. Reference Squire, Stark and Clark2004). Accordingly, in AD in the general population the early degeneration of the hippocampi and amygdalae is thought to underlie the memory impairments that are typical of (and occur early in) the disorder (e.g. Hyman et al. Reference Hyman, Van Hoesen and Damasio1990). However, within DS individuals with AD, we found no evidence that regional brain anatomy was significantly related to cognitive function.
One possible explanation for our failure to detect relationships between regional brain anatomy and cognitive function is that our study suffered from a lack of statistical power; that is, that we were limited by the size of our DS+AD sample. Alternatively, cognitive deficits associated with AD in DS individuals may be related to subtle differences in brain anatomy, detectable only using VBM. Another possibility is that cognitive deficits in this population may be more closely related to factors other than differences in brain anatomy. These may include, for example, amyloid plaque density (Wisniewski et al. Reference Wisniewski, Wisniewski and Wen1985), differences in brain metabolism and perfusion (Azari et al. Reference Azari, Horwitz, Pettigrew, Grady, Haxby, Giacometti and Schapiro1994) and/or other neurobiological factors, such as brain myo-inositol concentration (Beacher et al. Reference Beacher, Simmons, Daly, Prasher, Adams, Margallo-Lana, Morris, Lovestone, Murphy and Murphy2005). Further (and larger) studies are therefore required to relate other aspects of brain anatomy and metabolism to cognition in both non-demented and demented DS individuals.
Clinical utility of our findings
As this was a cross-sectional study, we could not determine the time course of changes, as opposed to differences, in brain anatomy. That would require a longitudinal approach. Therefore, it remains unclear whether changes in brain anatomy over time could be used to aid the clinical diagnosis of AD in people with DS. In addition, we do not suggest that any between-group differences we identified can be used alone as a diagnostic ‘test’. However, the diagnosis of AD in learning disabled populations is usually more difficult than in the general population. We found that that some of the significant between-group differences in brain anatomy were reliable ‘biomarkers’ for the presence of AD. Thus, in future, the combination of routine medical practice and brain imaging (if practicable) may help to build confidence in clinical diagnosis.
Summary
DS individuals with AD have significant reductions in medial temporal and striatal volumes, and these may provide markers of clinical AD.
Acknowledgements
This project was generously supported by the South London and Maudsley National Health Service (NHS) Trust (National Division) and the Baily Thomas Charitable Fund. We especially thank the individuals with Down's syndrome, and their families and carers, for taking part in the study.
Declaration of Interest
None.
