657 J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.73.6.657 on 1 December 2002. Downloaded from PAPER Detection of grey matter loss in mild Alzheimer’s disease with voxel based morphometry G B Frisoni, C Testa, A Zorzan, F Sabattoli, A Beltramello, H Soininen, M P Laakso ............................................................................................................................. J Neurol Neurosurg Psychiatry 2002;73:657–664 Objectives: To test the applicability of an automated method of magnetic resonance image analysis (voxel based morphometry) to detect presence and severity of regional grey matter density reduction—a proxy of atrophy—in Alzheimer’s disease. Methods: Twenty nine probable Alzheimer’s patients and 26 non-demented controls (mini-mental state examinations mean (SD) 21 (4) and 29 (1)) underwent high resolution 3D brain magnetic resonance imaging. Spatial normalisation to a stereotactic template, segmentation into grey matter, white matter, See end of article for and cerebrospinal fluid, and smoothing of the grey matter were carried out based on statistical para- authors’ affiliations ....................... metric mapping (SPM99) algorithms. Analyses were carried out: (a) contrasting all Alzheimer’s patients with all controls (p<0.05 corrected for multiple comparisons); (b) contrasting the three Alzheimer’s Correspondence to: patients with mini-mental state of 26 and higher with all controls (p<0.0001 uncorrected); and (c) cor- Dr G B Frisoni, Laboratory of Epidemiology and relating grey matter density with mini-mental state score within the Alzheimer’s group (p<0.0001 Neuroimaging, IRCCS San uncorrected). Giovanni di Dio-FBF, via Results: When all Alzheimer’s patients were compared with controls, the largest atrophic regions cor- Pilastroni 4, 25125 responded to the right and left hippocampal/amygdalar complex. All parts of the hippocampus (head, Brescia, Italy; body, and tail) were affected. More localised atrophic regions were in the temporal and cingulate gyri, [email protected] precuneus, insular cortex, caudate nucleus, and frontal cortex. When the mildest Alzheimer’s patients Received were contrasted with controls, the hippocampal/amygdalar complex were again found significantly 19 February 2002 atrophic bilaterally. The mini-mental state score correlated with grey matter density reduction in the In revised form 14 June temporal and posterior cingulate gyri, and precuneus, mainly to the right. 2002 copyright. Accepted 8 August 2002 Conclusions: Voxel based morphometry with statistical parametric mapping is sensitive to regional ....................... grey matter density reduction in mild Alzheimer’s disease. f the biological indicators of Alzheimer’s disease (AD), patients with mild to moderate AD. Moreover, the sensitivity to those detecting medial temporal lobe atrophy based on the very mild stages of AD will be assessed in those AD patients Othree dimensional T1 weighted magnetic resonance with highest mini-mental state score (26 and above).14 (MR) imaging are among the most sensitive. Hippocampal and entorhinal cortex volumetry and surface measures of the entorhinal cortex are believed to have high sensitivity and METHODS specificity in the detection of AD from non-demented elderly Subjects controls12 and can predict conversion of mild cognitive This study comprised 29 patients who met NINCDS-ADRDA http://jnnp.bmj.com/ 15 impairment to AD.3 However, the variability of the measure- criteria for probable AD and 26 non-demented controls ments because of the human tracer has so far limited direct recruited between September 1993 and December 1994 for a comparisons of the results of different research groups. The study on linear measures of brain atrophy in AD and have development of observer independent tools might be a signifi- been described in previous reports.16 AD were outpatients seen cant advancement. at the Alzheimer’s Unit, Brescia, Italy. Routine dementia To date, some computer based tools are available that can assessment including standardised history, laboratory exami- detect volume or shape changes in the regions of interest to nations, physical and neurological examination, neuropsycho- AD.4–7 Among these, one of the most user independent is sta- logical assessment, and routine computed tomography was on September 25, 2021 by guest. Protected tistical parametric mapping (SPM).8 Initially developed for carried out in all.16 Control subjects were patients’ relatives brain activation studies with positron emission tomography,9 (mostly spouses) without detectable cognitive deficit. All were it has been applied to different image analysis issues. In the given the mini-mental state examination (MMSE) and case of morphometry, SPM permits the comparison of the clinical dementia rating scale17 and were judged not to be density of grey matter on a voxel by voxel basis in a group of demented by a neurologist and a psychologist involved in the patients with that of a group of controls. The result of the evaluation of the patients. Uncontrolled physical diseases comparison is a three dimensional map of significant regional (heart or renal failure, cirrhosis, etc) were considered differences of grey matter density reduction, a close proxy of exclusion criteria, while risk factors such as hypertension and grey matter atrophy. The process is largely automatic and diabetes were accepted. Of the original 46 patients and 31 human intervention is comparatively low. Three studies have controls,16 17 and 4 could not be included in this study as some been published so far on AD,10–12 but inconsistent results, slices were missing. One additional control was excluded for different study populations and analysis protocols, and the present availability of more sophisticated image processing protocols13 argue for an independent confirmation. ............................................................. The aim of this study is to test the sensitivity of voxel based Abbreviations: AD, Alzheimer’s disease; MMSE, mini-mental state morphometry with SPM in the detection of grey matter density examination; SPM, statistical parametric mapping; MR, magnetic reduction and its association with global disease severity in resonance www.jnnp.com 658 Frisoni, Testa, Zorzan, et al J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.73.6.657 on 1 December 2002. Downloaded from ferent (p>0.09). Dementia severity was of mild to moderate Table 1 Sociodemographic and degree. All patients and controls were right handed. clinical features of 29 mild to moderate Three AD patients were in the highest decile of the MMSE AD patients and 26 non-demented distribution and were selected to investigate atrophy in very controls mild AD. Age, sex, and MMSE were: 66, F, 26; 81, F, 27; and 79, AD Controls M, 27. (n=29) (n=26) Written informed consent was obtained by both patients Age (y) and control subjects or their primary caregivers after mean (SD) 74 (9) 69 (8) discussion of risks and benefits of participation. No compen- range 53–86 54–86 sation was provided. The study was approved by the local eth- Women n (%) 23 (79) 17 (65) ics committee. Education (y) mean (SD) 7 (4) 8 (3) MR images acquisition range 2–18 5–19 For each subject, a high resolution sagittal T1 weighted volu- Mini Mental State Exam metric MR scan was acquired at the Radiology Department, mean (SD) 21 (4) 29 (1) University of Verona, using a 1.5 tesla Magnetom scanner range 12–27 25–30 (Siemens, Erlangen, Germany), with a gradient echo 3D tech- Clinical dementia rating scale n (%) nique: TR=10 ms, TE=4 ms, TI=300 ms, flip angle=10°, field 0 26 (100) of view=250 mm, acquisition matrix 160×256, slice thickness 0.5 8 (27.6) 1.3 mm, 128 slices. The same parameters were used for every 1 13 (44.8) subject. 2+ 8 (27.6) Image preprocessing After removing the voxels below the cerebellum with MRIcro (www.psychology.nottingham.ac.uk/staff/cr1/mricro.html),18 the development of multiple system atrophy a few years after MR scans were analysed with SPM99 (www.fil.ion.ucl.ac.uk/ scanning. spm) running under Matlab 6.0 (Mathworks, Sherborn, MA, Table 1 shows that patients were about five years older USA) on a Sun Sparc Ultra 30 workstation (Sun Microsystem, (p=0.04) than controls, while sex and education were not dif- Mountain View, CA). Before carrying out group comparisons, Table 2 Regional grey matter density reduction in 29 mild to moderate AD patients compared with 26 non-demented controls copyright. Stereotactic coordinates (mm) Cluster size k (mm3) Region xyzZ score 138 (1104) R hippocampus (head)/ amygdala 32 −4 −20 5.99 R hippocampus (head)/ amygdala 26 −6 −14 5.79 45 (360) R uncus 20 6 −22 5.92 R anterior amygdala 26 10 −26 5.84 45 (360) R hippocampus (tail) 24 −34 4 5.60 43 (344) R anterior cingulate 8 34 18 5.58 12 (96) L middle temporal gyrus −56 −14 −14 5.26 56 (448) L hippocampus (head) −26 −8 −16 5.21 13 (104) L hippocampus (body) −36 −28 −12 5.20 5 (40) L caudate (head) −10 20 2 5.14 http://jnnp.bmj.com/ 8 (64) R superior temporal gyrus 54 −42 14 5.09 33 (264) R middle temporal gyrus 52 −30 −2 5.07 R middle temporal gyrus 54 −22 −10 4.84 4 (32) L precuneus −10 −62 34 5.03 9 (72) R precuneus 18 −66 32 4.96 4 (32) L middle frontal gyrus −26 −10 56 4.95 4 (32) L middle frontal gyrus −28 44 −12 4.94 8 (64) R fusiform gyrus 52 −40 −22 4.89 3 (24) L precuneus −10 −54 36 4.82 1 (8) R hippocampus (body) 38 −24 −14 4.80 on September 25, 2021 by guest. Protected 1 (8) L insula −44 10 −2 4.78 1 (8) L anterior amygdala −24 8 −24 4.78 3 (24) R uncus 28 −8 −38 4.76 4 (32) R middle temporal gyrus 52 −8 −16 4.75 2 (16) R insula 44 −26 18 4.75 4 (32) R precuneus 4 −52 30 4.75 1 (8) R fusiform gyrus 58 −36 −26 4.75 2 (16) R insula 50 −4 8 4.74 1 (8) R hippocampus (tail) 38 −30 −12 4.74 1 (8) L posterior cingulate 0 −46 44 4.71 1 (8) R insula 38 4 12 4.71 1 (8) R hippocampus (tail) 36 −32 −10 4.70 1 (8) R superior temporal gyrus 54 −58 28 4.69 1 (8) L precuneus −30 −70 34 4.69 L=left, R=right.