Are Transversal MR Images Sufficient to Distinguish Persons with Mild Cognitive Impairment From Healthy Controls?

Are Transversal MR Images Sufficient to Distinguish Persons with Mild Cognitive Impairment From Healthy Controls? € nninghoff, MD, Martha Dlugaj, PhD, Oliver Kraff, PhD, Marie Henrike Geisel, MSc, Christoph Mo € ckel, PhD, Raimund Erbel, MD, Christian Weimar, MD, Isabel Wanke, MD on behalf of the Karl-Heinz Jo Heinz Nixdorf Recall Study Investigative Group Abbreviations AD Alzheimer disease aMCI amnestic mild cognitive impairment naMCI nonamnestic mild cognitive impairment CMB cerebral microbleeds FLAIR fluid-attenuated inversion recovery HNR Heinz Nixdorf Recall study MRI magnetic resonance imaging ROC receiver-operating characteristic rWTH radial width of the temporal horn of the lateral ventricle

Rationale and Objectives: Mild cognitive impairment (MCI) is associated with an increased risk of developing dementia. This study aims to determine whether current standard magnetic resonance imaging (MRI) is providing markers that can distinguish between subjects with amnestic MCI (aMCI), nonamnestic MCI (naMCI), and healthy controls (HCs). Materials and Methods: A subset of 126 MCI subjects and 126 age-, gender-, and educationappropriate HCs (mean age, 70.9 years) were recruited from 4157 participants in the longitudinal community-based Heinz Nixdorf Recall Study. The burden of white matter hyperintensities (WMHs), cerebral microbleeds, and brain atrophy was evaluated on transversal MR images from a single 1.5-T MR scanner by two blinded neuroradiologists. Logistic regression and receiver-operating characteristic analysis were used for statistical analysis. Results: Occipital WMH burden was significantly increased in aMCI, but not in naMCI relative to HCs (P = .01). The combined MCI group showed brain atrophy relative to HCs (P = .01) pronounced at caudate nuclei (P = .01) and temporal horn level (P = .004) of aMCI patients and increased at the frontal and occipital horns of naMCI patients compared to either aMCI or HCs. Microbleeds were equally distributed in the MCI and control group, but more frequent in aMCI (22 of 84) compared to naMCI subjects (3 of 23). Conclusions: In his cohort, increased occipital WMHs and cortical and subcortical brain atrophies at temporal horn and caudate nuclei level distinguished aMCI from naMCI subjects and controls. Volumetric indices appear of interest and should be assessed under reproducible conditions to gain diagnostic accuracy. Key Words: mild cognitive impairment; 1.5 Tesla MRI; ageing and cognition; brain atrophy; white matter hyperintensities. ªAUR, 2015

SD standard deviation WMH white matter hyperintensities

Acad Radiol 2015; 22:1172–1180 From the Institute for Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, University of Duisburg-Essen, Essen, Germany (C.M., O.K., I.W.); Department of Neurology, University Hospital Essen (M.D., C.W.), Erwin L. Hahn Institute for Magnetic Resonance Imaging (O.K.), Institute for Medical Informatics, Biometry and Epidemiology (M.H.G., K.-H.J.), Clinic of Cardiology, West-German Heart

1172

Centre, University Hospital Essen (R.E.), University of Duisburg-Essen, Essen, Germany; and Neuroradiology, Swiss Neuro Institute, Clinic Hirslanden, Zurich, Switzerland (I.W.). Received November 25, 2014; accepted April 15, 2015. Address correspondence to: C.M. e-mail: [email protected] ªAUR, 2015 http://dx.doi.org/10.1016/j.acra.2015.04.008

Academic Radiology, Vol 22, No 9, September 2015

TRANSVERSAL MR IMAGES TO DIAGNOSE MILD COGNITIVE IMPAIRMENT?

M

ild cognitive impairment (MCI) describes a transitional state between cognitive changes of normal aging and dementia, especially Alzheimer disease (AD) (1,2). Hence, MCI is a risk factor for dementia with an estimated conversion rate of 10%–15% per year, compared to 1%–2% in the cognitively normal, elderly population (3). This prodromal AD state is diagnosed in the clinical setting by neurologic and neuropsychological assessment (1,2,4). Nonamnestic forms of MCI (naMCI) have shown findings related to vascular disease, whereas amnestic MCI (aMCI) subjects demonstrated demographic, genetic, and magnetic resonance imaging (MRI) characteristics similar to AD pathology (2,5). Despite a controversial definition of MCI as a diagnostic entity, because it does not constitute a homogeneous clinical syndrome, it is an ideal target for prevention and future therapies of dementia (3,6). An early diagnosis and differentiation of MCI subtypes may have a major impact on the selection of suitable prospective therapies in future. MRI examinations are not routinely included in clinical work-up of MCI, although brain tissue and brain volumetric changes may help to predict conversion from MCI to dementia (7,8). Cerebral microbleeds (CMBs) are often in focus of MR studies dealing with neurodegenerative diseases and dementia, but their predictive value in MCI or AD-converters remains unclear (9–13). Medial temporal lobe atrophy has been shown to be an important predictor for conversion from MCI to AD (14). The volume of the hippocampus, the entorhinal cortex, and amygdala is known to decline in early stages of AD (15,16). We hypothesized that the degrees of brain atrophy and the burden of white matter hyperintensities (WMHs) and CMBs on transversal 1.5-T MR images correlate with the clinical diagnosis of MCI. Thus, we compared MR signs in subjects with MCI (divided into amnestic and nonamnestic subtypes because of the different underlying etiology) with age-, gender-, and education-matched controls on transversal fluid-attenuated inversion recovery (FLAIR) and T2*-weighted images in a large German population-based study. This study aims to determine whether current standard MRI is providing markers that can distinguish between subjects with aMCI, naMCI, and HCs in a single study. MATERIALS AND METHODS Study Population and Sampling Procedure

The Heinz Nixdorf Recall (HNR; Risk Factors, Evaluation of Coronary Calcium and Lifestyle) study is a populationbased prospective cohort study with 4814 subjects (age range, 45–75 years) randomly selected from mandatory lists of residence in the Ruhr area in Germany (17). The major aim of the HNR study was to evaluate the predictive value of coronary artery calcification using electron-beam computed tomography for myocardial infarction and cardiac death in comparison to cardiovascular risk factors. Study methods have been described elsewhere in detail (17).

Five years after baseline examination, a follow-up was performed (response rate, 90.5%; n = 4359) and combined with a substudy to determine the prevalence and risk factors of cognitive impairment (particularly MCI, its subtypes, and AD) in this cohort. Hence, a cognitive screening battery including memory testing was included (18). Sampling methods are described elsewhere in detail (18). A neuropsychologist performed the standardized neuropsychological examination using the Alzheimer’s Disease Assessment Scale, the number connection test of the N€ urnberg Gerontopsychological Inventory, the verbal fluency test, and an Instrumental Activities of Daily Living scale (18). Participants with dementia, severe depression, Parkinson disease, mental retardation, severe alcohol consumption, known brain cancer, severe problems with the German language, and severe auditory impairment leading to invalid cognitive testing were excluded. After exclusion of study participants with MRI contraindications, this study included a subsample of 252 (85%) subjects, namely 126 MCI subjects and 126 matched controls. All participants gave their written informed consent before neuropsychological and MRI examinations. This substudy was approved by the responsible Ethics Committee of our University Hospital and followed established guidelines on good epidemiologic practice. Diagnostic Classifications and Covariates

MCI was diagnosed according to the current International Working Group on MCI criteria with exception of the complaint criterion (subject or informant had to express some concern about the subject’s cognitive function) (2,4). According to the different underlying etiology, we distinguished aMCI, presenting an objective impairment (always defined as a performance one standard deviation [SD] below the mean) in memory with or without impairment in any other cognitive domain, from naMCI, if a cognitive domain other than memory was impaired (2,4). Study participants without impairment of any cognitive domain were assessed as HCs. The covariate of education was classified by the International Standard Classification of Education as total years of formal education, combining school and vocational training (19). MR Imaging Techniques and Examination Protocols

All study participants diagnosed with MCI and their matched controls were examined on a single 1.5-T MR scanner (Magnetom Avanto, Siemens Healthcare, Erlangen) of 160-cm length, with 60-cm bore diameter, 40/40/45 mT/ m gradient strength, and 200 T/m/s slew rate. The MR scanner was equipped with a 12-channel receive-only matrix head coil provided by the vendor. A transversal FLAIR sequencing (Repetition time [TR] = 9000 ms, echo time [TE] = 109 ms, flip angle = 150 , voxel size = 0.9  0.9  6.0 mm, bandwidth = 130 Hz/Px, and acquisition time = 3:38 minutes) was performed to detect white matter lesions and 1173

€ MONNINGHOFF ET AL

to assess subcortical brain atrophy. A transversal T2*-weighted gradient echo sequence (TR = 872 ms, TE = 26 ms, flip angle = 20 , voxel size = 1.0  1.0  1.5 mm, bandwidth = 80 Hz/Px, parallel acquisition technique factor 2 [GRAPPA], 2 averages, and acquisition time = 9:25 minutes) in axial orientation was used to identify CMBs. MR Image Assessment

Quantitative and qualitative MR image assessment was performed by two neuroradiologists (C.M., I.W.) in consensus and blinded to the identity and neuropsychological diagnosis of the subjects. Minor motion artifacts were present in the data sets of 44 (17%) of 252 subjects, but all acquired MR images were qualitatively suitable for further analysis. Lesion counts and measurements of the ventricle-to-brain ratio were performed manually on a picture archiving and communication system console by both neuroradiologists. WMHs were assessed on axial FLAIR images and subdivided into periventricular and subcortical lesions corresponding to the Breteler scale (20). Periventricular WMHs were defined as bright spots with direct contact to the ventricle and scored in following locations: frontal caps (adjacent to frontal horns), lateral bands (adjacent to lateral wall of the ventricles), and occipital caps (adjacent to the occipital horns) and classified in three stages: none or pencil thin, smooth halo, and large confluent white matter lesions (21). Subcortical WMHs were defined as bright isolated or confluent spots in the white matter without contact to ventricle walls or periventricular WMHs. Frontal, parietal, temporal, and occipital lesions of both hemispheres were differentiated into three groups: 1–3 mm, >3–10 mm, and >10 mm. Cortical atrophy was rated visually on axial FLAIR images using a four-point severity scale, which discriminates none, low, moderate, and severe atrophy (20). Subcortical brain atrophy was estimated by the use of the ventricle-to-brain ratio at frontal horn level, caudate nucleus level, and occipital horn level. Hippocampal atrophy typical of AD primarily occurs in the region of the hippocampal head. In this study, hippocampal atrophy was assessed by the radial width of the temporal horns of the lateral ventricles (rWTH), which was elsewhere described as a sensitive marker for AD (22). The rWTH was measured at the tip of the horns on axial, 6-mm thick FLAIR images on both sides separately by one observer (C.M.) to capture a possible asymmetry of temporal brain atrophy. Microbleeds were defined as homogeneous, dot-like signal loss lesions of #10-mm size in gray and white matter on T2*-weighted images. Their number and location were registered for all study participants.

Academic Radiology, Vol 22, No 9, September 2015

by MCI subgroups (amnestic vs. nonamnestic). We used logistic regression conditioned on matched pairs to compare MCI subjects with controls and also stratified by MCI group. Logistic regression (adjusted for age) was used to compare MCI subgroups. Fisher’s exact test was used to compare aMCI and naMCI regarding CMB prevalence. A receiver-operating characteristic (ROC) analysis was performed for all continuous variables with statistically significant differences between the MCI and control group using SPSS software (IBM SPSS statistics for Windows, version 22.0, Armonk, NY). All other analyses were performed using SAS software (version 9.2, SAS Institute Inc., Cary, NC). All reported P values are nominal, two-sided, and based on likelihood ratio tests. A significance level a of 5% was applied. RESULTS Descriptive Statistics

The sociodemographic characteristics are presented in Table 1 for study participants, stratified by controls, MCI subjects, and diagnostic subgroups. aMCI was diagnosed in 93 (74%) and naMCI was diagnosed in 33 (26%) subjects. aMCI study participants were significantly older than naMCI study participants (mean age, 72.3 years vs. 66.9 years, P < .0001) with a surplus of male subjects (62.4% vs. 48.5%, P < .17). There were no significant differences between both MCI subgroups regarding the marital status (P = .30), education (P = .22), hypertension (P = .24), and coronary heart disease (P = .75). White Matter Hyperintensities

WMHs on transversal FLAIR images were subdivided into periventricular and subcortial hyperintensities (Fig. 1). Frontal and lateral periventricular WMHs were equally distributed in aMCI and naMCI subjects as well as in MCI and control subjects (Table 2). However, occipital periventricular WMHs were significantly larger in study participants diagnosed with aMCI (P = .01) and in all MCI subjects compared to controls (P = .02). Large confluent occipital caps were found in 11 (12%) of 93 persons with aMCI, but only in 2 (6%) of 33 persons with naMCI and in 4 (3%) of 126 control subjects. In MCI subjects, parietal subcortial WMHs of 1- to 3-mm size (P = .03) and >3- to 10-mm size (P = .01) were more frequent than in the control group (data not shown). In all other brain regions, no statistically significant differences were found between the three subgroups of study participants regarding the number and location of subcortial WMHs (data not shown).

Statistical Methods

Cortical and Subcortical Brain Atrophy

Continuous variables are summarized as mean  SD. Discrete variables are reported as absolute frequencies and in percent. Distributions of sociodemographic and clinical characteristics were assessed for controls and MCI subjects and also stratified

The estimated cortical brain atrophy was significantly higher in MCI subjects (P = .01; Table 3) and in aMCI subjects (P = .03) compared to matched controls. Severe cortical brain atrophy was found in 9 of 93 (10%) aMCI, in 2 of 33 (6%)

1174

Academic Radiology, Vol 22, No 9, September 2015

TRANSVERSAL MR IMAGES TO DIAGNOSE MILD COGNITIVE IMPAIRMENT?

TABLE 1. Characteristics of Subjects with MCI and Controls and Stratified by the Two MCI Subgroups

Parameter Age, y, mean  SD Gender (%) Male Female Education (%) #10 y 11–13 y $14 y Marital status (%) Single Married Divorced or in separation Widowed CHD (%) Yes No Hypertension (JNC7) None Stage 1 or 2

P Value, MCI Versus Controls

MCI, Amnestic (n = 93)

MCI, Nonamnestic (n = 33)

70.8  6.4

Matched

72.3  5.6

66.9  7.0

74 (59) 52 (41)

74 (59) 52 (41)

Matched

58 (62) 35 (38)

16 (48) 17 (52)

.17

28 (22) 80 (64) 18 (14)

21 (17) 75 (60) 30 (24)

Matched

17 (18) 62 (67) 14 (15)

11 (33) 18 (55) 4 (12)

.22

2 (2) 87 (70) 10 (8)

7 (6) 91 (72) 8 (6)

2 (2) 63 (68) 6 (7)

0 (0) 24 (75) 4 (139)

25 (20)

20 (16)

21 (23)

4 (13)

.30

18 (14) 108 (86)

17 (13) 109 (87)

.86

12 (13) 81 (87)

5 (15) 28 (85)

.75

>.99

53 (57) 40 (43)

22 (69) 10 (31)

.24

MCI Total (n = 126) 70.9  6.4

76 (60) 50 (40)

Controls (n = 126)

Matched

75 (60)* 50 (40)

P Value, aMCI Versus naMCI <.0001

aMCI, amnestic MCI; CHD, coronary heart disease; JNC7, hypertension defined according to ‘‘The Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure’’; MCI, mild cognitive impairment; naMCI, nonamnestic MCI; SD, standard deviation. Data are presented as mean age (standard deviation) and number (percentage) of study participants. Significant P values are presented in bold. *Data of one study participant were not available.

Figure 1. Fluid-attenuated inversion recovery images of a 79-year-old female amnestic mild cognitive impairment subject (right) with large confluent white matter hyperintensities (WMHs) around both lateral ventricles and with low cortical brain atrophy compared to her matched control (left) without WMHs and without significant cortical brain atrophy.

naMCI, and in 2 of 126 (2%) control subjects. Moderate-tosevere cortical brain atrophy was more common in MCI subjects (33%) than in cognitively unimpaired controls (19%). No brain atrophy was found in 36% of the MCI subjects and in 47% of the control subjects.

Semiquantitative assessment of subcortical brain atrophy was performed and the resulting ventricle-to-brain ratio was calculated (Table 4). A high ventricle-to-brain ratio indicates subcortical brain atrophy due to enlarged ventricles and thinning of adjacent brain parenchyma. Subcortical atrophy at 1175

€ MONNINGHOFF ET AL

Academic Radiology, Vol 22, No 9, September 2015

TABLE 2. Periventricular WMHs in MCI Subjects (Total and Stratified by Subgroups) and Matched Controls Subdivided in 3 Periventricular Regions and 4 Size Classes

Periventricular WMHs

MCI Total (n = 126)

Controls (n = 126)

95 (75) 27 (22) 4 (3)

95 (75) 26 (21) 5 (4)

91 (72) 31 (25) 4 (3)

96 (76) 28 (22) 2 (2)

24 (19) 17 (14) 13 (10)

28 (22) 9 (7) 4 (3)

Frontal None/pencil thin Smooth halo Confluent Lateral None/pencil thin Smooth halo Confluent Occipital None/pencil thin Smooth halo Confluent

P Value

Amnestic MCI (n = 93)

Nonamnestic MCI (n = 33)

P Value, aMCI Versus naMCI (Adj.)

P Value, aMCI Versus Controls

P Value, naMCI Versus Controls

.97

70 (76) 19 (20) 4 (4)

25 (76) 8 (24) 0 (0)

.20

.99

.68

.74

65 (70) 24 (26) 4 (4)

26 (79) 7 (21) 0 (0)

.29

.81

.72

.02

20 (22) 14 (15) 11 (12)

4 (12) 3 (9) 2 (6)

.55

.01

.90

Adj, age adjusted; aMCI, amnestic MCI; MCI, mild cognitive impairment; naMCI, nonamnestic MCI; WMH, white matter hyperintensity. Data are presented as number of study participants (percentage). Significant P values are presented in bold.

TABLE 3. Brain Atrophy on Axial FLAIR Images Estimated on a Four-Point Scale

Brain Atrophy Estimation None Low Moderate Severe

MCI Total (n = 126)

Matched Controls (n = 126)

46 (36) 39 (31) 30 (24) 11 (9)

59 (47) 43 (34) 22 (17) 2 (2)

P Value

Amnestic MCI (n = 93)

Nonamnestic MCI (n = 33)

P Value, aMCI Versus naMCI (Adj.)

P Value, aMCI Versus Controls

P Value, naMCI Versus Controls

.01

27 (29) 32 (34) 25 (27) 9 (10)

19 (58) 7 (21) 5 (15) 2 (6)

.79

.03

.33

Adj, age adjusted; aMCI, amnestic MCI; FLAIR, fluid-attenuated inversion recovery; MCI, mild cognitive impairment; naMCI, nonamnestic MCI. Data are presented as number (percentage) of study participants. Significant P values are presented in bold.

frontal horn level was almost equal in the diagnostic subgroups with apparently smaller brain and ventricle diameters in the naMCI group. The occipital ventricle-to-brain ratio was almost equal in aMCI and HCs (P = .53), but reduced in naMCI, which distinguished aMCI from naMCI subjects (P = .049). At caudate nuclei level, subcortical atrophy was significantly pronounced in all MCI and aMCI subjects compared to HCs (P = .01). A ventricle-to-brain ratio at caudate nuclei level of 0.5 (cutoff point) fails to separate MCI from cognitively healthy subjects because of a sensitivity of 50% and a specitivity of 68% (area under the ROC curve [AUC], 0.585; standard error, 0.036; and 95% confidence interval [CI], 0.514–0.655). The ventricle and brain diameters at frontal and occipital horn level were apparently smaller in the naMCI group compared to the aMCI group. The rWTH was significantly increased in the MCI group compared to matched HCs (right: P = .0035, OR, 1.323; CI 95%, 1.096–1.596; left: P = .0046, OR, 1.303; CI 95%, 1.085–1.564) with emphasized atrophy of the left side (Table 4). In the aMCI subgroup, the rWTH was also significantly higher compared to HCs (right: P = .0065, OR, 1.342; 95% CI, 1.086–1.657; left: P = .0131, OR, 1.292; 1176

95% CI, 1.055–1.582). However, the ROC analysis revealed poor sensitivity (56%) and specitivity (64%) for 5.9-mm rWTH (cutoff point) between MCI and control subjects on both sides (eg, right side: AUC, 0.592; standard error, 0.36; 95% CI, 0.522–0.663). In summary, aMCI was found to be associated with subcortical brain atrophy at caudate nuclei level. At occipital horn level, subcortical brain atrophy distinguished aMCI from naMCI and controls. The rWTH as a marker for the mediotemporal brain atrophy and the ventricle-to-brain ratio at caudate nuclei level as a marker for subcortical brain atrophy were significantly higher in the MCI group compared to the control group; however, the accuracy of both markers was useless to separate MCI from control subjects in our community-based case–control study. Cerebral Microbleeds

T2*-weighted images were available in 232 (92%) of 252 study participants to identify CMBs (Fig. 2). A total number of 99 (range, 1–34) CMBs were identified in 25 MCI subjects and a total number of 82 (range, 1–34) microbleeds in 22

Academic Radiology, Vol 22, No 9, September 2015

TRANSVERSAL MR IMAGES TO DIAGNOSE MILD COGNITIVE IMPAIRMENT?

TABLE 4. Ventricle and Brain Diameter, Ventricle-to-Brain Ratio at Three Levels, and rWTH to Assess Subcortical Brain Atrophy in aMCI, naMCI, and Matched Controls

Subcortical Brain Atrophy, mm, Mean  SD

MCI Total (n = 126)

Matched Controls (n = 126)

Frontal horn level Ventricle 36.35  5.71 37.98  7.49 Brain 106.07  10.56 107.53  8.20 Ventricle/brain ratio 0.34  0.05 0.37  0.23 Occipital horn level Ventricle 61.32  8.35 63.41  6.18 Brain 125.88  12.78 128.36  6.09 Ventricle/brain ratio 0.48  0.06 0.49  0.04 Caudate nuclei level Ventricle 18.38  4.57 17.83  3.82 Brain 109.79  6.59 112.51  6.42 Ventricle/brain ratio 0.17  0.04 0.16  0.03 rWTH Right 4.33  1.72 3.75  1.20 Left 4.18  1.73 3.60  1.25

P Value

Amnestic MCI (n = 93)

Nonamnestic MCI (n = 33)

P Value, P Value, P Value, aMCI naMCI aMCI Versus Versus Versus naMCI (Adj.) Controls Controls

.03 .19 .09

37.26  4.89 33.79  7.04 107.38  4.34 102.39  19.04 0.35  0.04 0.32  0.07

.03 .01 .11

.17 .74 .23

.04 .09 .07

.02 .02 .07

62.54  6.41 57.88  11.74 127.20  5.72 122.15  22.90 0.49  0.05 0.46  0.09

.01 .01 .049

.20 .14 .53

.02 .07 .04

.24 .0001 .01

19.16  4.34 16.18  4.54 109.81  6.67 109.76  6.43 0.17  0.04 0.15  0.03

.09 .33 .30

.14 <.0001 .01

.88 .64 .64

.56 .85

.01 .01

.30 .17

.004 .005

4.44  1.88 4.33  1.90

4.01  1.10 3.76  1.14

Adj, age adjusted; aMCI, amnestic MCI; MCI, mild cognitive impairment; naMCI, nonamnestic MCI; rWTH, radial width of the temporal horn; SD, standard deviation. Data are presented as mean values in millimeter (standard deviation). Significant P values are presented in bold.

Figure 2. Cerebral microbleeds (arrows) located in the occipital cortex of a 75-year-old male mild cognitive impairment subject (right) and normal finding of the matched control (left) on axial T2*-weighted images.

control subjects. CMB were more frequently found in aMCI (22 of 83, 27%) compared to naMCI subjects (3 of 24, 13%). Regarding location and amount of CMB, no significant differences were evident between MCI subjects and their matched controls (Table 5).

DISCUSSION We aimed to correlate the burden of WMHs, brain atrophy, and CMBs on transversal FLAIR and T2*-weighted images

with the neuropsychological diagnoses of aMCI, naMCI, and HCs to identify MR characteristics as possible biomarkers. The motivation to confine this study to transversal MR images is based on the fact that most routine MR examinations contain these sequences. The use of threedimensional (3D) MR sequences for brain volumetry is state of the art to evaluate especially medial temporal lobe atrophy, which has been shown to be the most important predictive factor for conversion from MCI to AD (14,23). In the clinical setting, the diagnosis MCI is primarily based on clinical and cognitive evaluation without MRI 1177

€ MONNINGHOFF ET AL

Academic Radiology, Vol 22, No 9, September 2015

TABLE 5. Cerebral Microbleeds in MCI Subjects and Matched Controls and Stratified for the MCI Subgroups

Cerebral Microbleeds n (%) Cortex Frontal Parietal Temporal Occipital White matter Frontal Parietal Temporal Occipital Basal ganglia Thalamus Brainstem Cerebellum

MCI n = 108

Controls n = 125

P Value, MCI Versus Controls

aMCI n = 83

naMCI n = 24

P Value, aMCI Versus naMCI

3 (3) 6 (6) 3 (3) 1 (1)

5 (4) 6 (5) 2 (2) 3 (2)

.71 .76 .66 .34

2 (2) 6 (7) 3 (4) 1 (1)

1 (4) 0 (0) 0 (0) 0 (0)

.77 .41 .65 >.99

3 (3) 2 (2) 4 (4) 2 (2) 7 (6) 3 (3) 3 (3) 4 (4)

5 (4) 3 (2) 2 (2) 1 (1) 3 (2) 1 (1) 0 (0) 9 (7)

.48 .66 .42 .57 .22 .34 .99 .37

3 (4) 2 (3) 3 (4) 2 (2) 6 (7) 3 (4) 3 (4) 3 (4)

0 (0) 0 (0) 1 (4) 0 (0) 1 (4) 0 (0) 0 (0) 1 (4)

>.99 >.99 .38 .69 .23 .42 .12 .65

aMCI, amnestic MCI; MCI, mild cognitive impairment; naMCI, nonamnestic MCI. Data are presented as number of microbleeds (percentage). Significant P values are presented in bold. No T2*-weighted images were available in 18 MCI (9 aMCI/9 naMCI) subjects and in 2 control subjects.

examinations. This judgment allows to identify symptomatic, but nondemented individuals and to stratify them into subgroups with different probabilities regarding the underlying etiology: aMCI is more likely to progress to AD, whereas naMCI is more likely to progress to non-AD dementia (3). Hence, a current research interest is the identification of biomarkers that determine the underlying pathophysiology and the likelihood of progression for an individual MCI subject (24). Recent studies and reviews focus on multimodal MRI techniques to define surrogate neuroimaging biomarkers, which help to differentiate individuals with cognitive decline due to normal aging, MCI, or AD (6–8,24–26). WMHs on FLAIR images are a common finding among elderly cognitively healthy subjects, in those with MCI and a variety of dementias (27). The appearance of WMHs is attributed to processes of normal aging, cerebrovascular disease, and neurodegenerative processes, such as gliosis, microglial infiltration, and amyloid angiopathy (28). A long-term longitudinal study has shown that the WMH volume increases rapidly in normal subjects who develop MCI a decade later, suggesting that these lesions might be useful MR characteristics of MCI (29). Our study participants diagnosed with MCI had significantly larger occipital periventricular WMHs compared to cognitively unimpaired controls (P = .02). Other study groups found a significant correlation between parietal WMH and executive function decline in MCI or an association of a high lesion load in the temporal region with the risk of developing MCI, respectively (30,31). Several longitudinal MR studies have described brain atrophy patterns in MCI

1178

subjects, which should predict conversion from MCI to AD (25,32–34). Medial temporal lobe atrophy and hippocampal volume loss have been shown to be the best morphologic predictors for conversion from MCI to AD by volumetric MR studies (35–37). Measurements of the linear width of the temporal lobe and the rWTH have shown to be useful markers to differentiate AD subjects from controls (22,38–40). In contrast to other studies dealing with AD patients and CT scans, we applied the latter method to patients diagnosed with MCI. However, we could not verify the high ability of the rWTH to distinguish MCI from HCs in our study population using conventional axial FLAIR images. Another study group found good correlation of WMH scores and medial temporal atrophy, which were both increased in subjects with probable AD compared to cognitively unimpaired and naMCI subjects (27). MCI and especially aMCI subjects of our study population presented with pronounced cortical atrophy compared to HCs. Subcortical brain volume loss at caudate nucleus level was apparent in MCI subjects, especially in aMCI subjects, compared to their matched controls (Table 4). However, the accuracy of this marker was too low to distinguish between both groups. Subcortical white matter changes were not suitable to distinguish between the MCI subgroups and matched controls. The ventricle-to-brain ratio at frontal and occipital horn level in the Rotterdam Scan Study ranged from 0.21 to 0.45, including 1077 persons from the general population and aged 60–90 years (41). At both levels, our MCI control study population reached fractional ratios between 0.32 and 0.49, which indicate pronounced subcortical atrophy (Table 4).

Academic Radiology, Vol 22, No 9, September 2015

TRANSVERSAL MR IMAGES TO DIAGNOSE MILD COGNITIVE IMPAIRMENT?

CMBs were described as a significant risk factor for progressive cognitive impairment in subjects with MCI because of their association with microvascular disease, especially lacunar infarctions and increased WMH scores (9,11,13). We identified CMB in 23% of MCI and 18% in cognitively not impaired control subjects on T2*-weighted images. New imaging modalities like susceptibility-weighted imaging (SWI) at 7-T MRI identified CMB in 38% of MCI participants or in 78% of MCI/AD patients, but only in 44% of HCs (10,13). Limitations

Semiquantitative evaluation of brain atrophy was performed by two blinded neuroradiologists in consensus, which does not allow an estimation of inter-rater variability. This study is limited by the use of anisotropic MR data, which are mostly acquired in clinical practice, but not suitable for brain volumetry, for example, of the hippocampus. Medial temporal lobe atrophy was indirectly assessed by use of the rWTH on axial T2-weighted FLAIR images. Meanwhile, another study has shown that radiologic assessment of hippocampal atrophy is as good as computer-based volumetry for the classification of AD, MCI nonconverter, and normal controls and less good for the classification of MCI converter versus HCs (42). Our approach, taking into account the localization and count of WHMs, CMBs and general as well as mediotemporal brain atrophy could not define useful cutoff points for the selected MR markers to distinguish accurately between MCI and control subjects. Ongoing studies, including additionally acquired 3D MR data sets at baseline, after 28 months and after 5 years will provide more detailed information on longitudinal brain volume and tissue changes of this study population. SWI sequences were found to be superior to T2*-weighted gradient echo sequences regarding the detection of microbleeds after the starting point of our study. Meanwhile, SWI is widely used in clinical practice, but we did not change this sequence for reasons of uniformity. CONCLUSIONS In our community-based study population, subcortical brain atrophy at caudate nuclei level, general brain atrophy on a 4-point scale, occipital periventricular WMHs, and the rWTH were significantly higher in the MCI and aMCI groups compared to HCs. CMBs and WMH burden of most brain regions were not suitable to distinguish between aMCI, naMCI, and HCs. In conclusion, the applied semiquantitative assessment of brain atrophy and WMH burden on transversal FLAIR images is insufficient to differentiate accurately between MCI and control subjects in clinical routine. Recent studies suggest a combination of neuropsychological diagnostics, multimodal neuroimaging including brain volumetry of 3D MR sequences and cerebrospinal fluid markers to distinguish between cognitively normal adults, aMCI, and AD patients with high diagnostic accuracy (8,43).

ACKNOWLEDGMENTS The authors thank the Heinz Nixdorf Foundation for the generous support of the study. The study is also supported by the German Ministry of Education and Science. The authors thank all participating radiologic technicians, who contributed to the performance of MR examinations and who cared for study participants. The specific part of cognitive screening was funded by the German Research Council (DFG; SI236/10-1). For the MCI substudy, the authors received a research grant from the Dr. Werner Jackst€adtStiftung. The authors are indebted to the investigative group, the study personnel, and all study participants. The authors have no conflict of interest to disclose and the funders had no role in design and conduction of the study; collection, management, analysis, and interpretation of data; or preparation, review, or approval of the article. This study was approved by the Ethics Committee of the medical faculty, University Hospital Essen, and followed established guidelines on good epidemiological practice and has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments.

REFERENCES 1. Petersen RC, Smith GE, Waring SC, et al. Mild cognitive impairment: clinical characterization and outcome. Arch Neurol 1999; 56(3):303–308. 2. Winblad B, Palmer K, Kivipelto M, et al. Mild cognitive impairment–beyond controversies, towards a consensus: report of the International Working Group on Mild Cognitive Impairment. J Intern Med 2004; 256(3):240–246. 3. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med 2004; 256(3):183–194. 4. Petersen RC, Doody R, Kurz A, et al. Current concepts in mild cognitive impairment. Arch Neurol 2001; 58(12):1985–1992. 5. He J, Farias S, Martinez O, et al. Differences in brain volume, hippocampal volume, cerebrovascular risk factors, and apolipoprotein E4 among mild cognitive impairment subtypes. Arch Neurol 2009; 66(11):1393–1399. 6. Albert MS, DeKosky ST, Dickson D, et al. The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 2011; 7(3): 270–279. 7. Heister D, Brewer JB, Magda S, et al. Predicting MCI outcome with clinically available MRI and CSF biomarkers. Neurology 2011; 77(17):1619–1628. 8. Lin AL, Laird AR, Fox PT, et al. Multimodal MRI neuroimaging biomarkers for cognitive normal adults, amnestic mild cognitive impairment, and Alzheimer’s disease. Neurol Res Int 2012; 2012:907409. 9. Schneider JA. Brain microbleeds and cognitive function. Stroke 2007; 38(6):1730–1731. 10. Brundel M, Heringa SM, de Bresser J, et al. High prevalence of cerebral microbleeds at 7Tesla MRI in patients with early Alzheimer’s Disease. J Alzheimers Dis 2012. 11. Ayaz M, Boikov AS, Haacke EM, et al. Imaging cerebral microbleeds using susceptibility weighted imaging: one step toward detecting vascular dementia. J Magn Reson Imaging 2010; 31(1):142–148. 12. Cordonnier C, van der Flier WM, Sluimer JD, et al. Prevalence and severity of microbleeds in a memory clinic setting. Neurology 2006; 66(9): 1356–1360. 13. Kirsch W, McAuley G, Holshouser B, et al. Serial susceptibility weighted MRI measures brain iron and microbleeds in dementia. J Alzheimers Dis 2009; 17(3):599–609. 14. Korf ES, Wahlund LO, Visser PJ, et al. Medial temporal lobe atrophy on MRI predicts dementia in patients with mild cognitive impairment. Neurology 2004; 63(1):94–100. 15. Chupin M, Gerardin E, Cuingnet R, et al. Fully automatic hippocampus segmentation and classification in Alzheimer’s disease and mild cognitive

1179

€ MONNINGHOFF ET AL

16.

17.

18.

19. 20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

impairment applied on data from ADNI. Hippocampus 2009; 19(6): 579–587. Laakso MP, Soininen H, Partanen K, et al. Volumes of hippocampus, amygdala and frontal lobes in the MRI-based diagnosis of early Alzheimer’s disease: correlation with memory functions. J Neural Transm Park Dis Dement Sect 1995; 9(1):73–86. Schmermund A, Mohlenkamp S, Stang A, et al. Assessment of clinically silent atherosclerotic disease and established and novel risk factors for predicting myocardial infarction and cardiac death in healthy middle-aged subjects: rationale and design of the Heinz Nixdorf RECALL Study. Risk Factors, Evaluation of Coronary Calcium and Lifestyle. Am Heart J 2002; 144(2):212–218. Dlugaj M, Weimar C, Wege N, et al. Prevalence of mild cognitive impairment and its subtypes in the Heinz Nixdorf Recall study cohort. Dement Geriatr Cogn Disord 2010; 30(4):362–373. UNESCO. International standard classification of education (ISCED), 1997. de Groot JC, de Leeuw FE, Oudkerk M, et al. Cerebral white matter lesions and cognitive function: the Rotterdam Scan Study. Ann Neurol 2000; 47(2): 145–151. Mantyla R, Erkinjuntti T, Salonen O, et al. Variable agreement between visual rating scales for white matter hyperintensities on MRI. Comparison of 13 rating scales in a poststroke cohort. Stroke 1997; 28(8):1614–1623. Frisoni GB, Geroldi C, Beltramello A, et al. Radial width of the temporal horn: a sensitive measure in Alzheimer disease. AJNR Am J Neuroradiol 2002; 23(1):35–47. Duara R, Loewenstein DA, Potter E, et al. Medial temporal lobe atrophy on MRI scans and the diagnosis of Alzheimer disease. Neurology 2008; 71(24):1986–1992. Aksu Y, Miller DJ, Kesidis G, et al. An MRI-derived definition of MCI-to-AD conversion for long-term, automatic prognosis of MCI patients. PLoS One 2011; 6(10):e25074. Spulber G, Niskanen E, MacDonald S, et al. Whole brain atrophy rate predicts progression from MCI to Alzheimer’s disease. Neurobiol Aging 2010; 31(9):1601–1605. Wolz R, Julkunen V, Koikkalainen J, et al. Multi-method analysis of MRI images in early diagnostics of Alzheimer’s Disease. PLoS One 2011; 6(10): e25446. Appel J, Potter E, Bhatia N, et al. Association of white matter hyperintensity measurements on brain MR imaging with cognitive status, medial temporal atrophy, and cardiovascular risk factors. AJNR Am J Neuroradiol 2009; 30(10):1870–1876. Du AT, Schuff N, Chao LL, et al. White matter lesions are associated with cortical atrophy more than entorhinal and hippocampal atrophy. Neurobiol Aging 2005; 26(4):553–559. Silbert LC, Dodge HH, Perkins LG, et al. Trajectory of white matter hyperintensity burden preceding mild cognitive impairment. Neurology 2012; 79(8):741–747.

1180

Academic Radiology, Vol 22, No 9, September 2015

30. Jacobs HI, Visser PJ, Van Boxtel MP, et al. Association between white matter hyperintensities and executive decline in mild cognitive impairment is network dependent. Neurobiol Aging 2012; 33(1):201, e1-8. 31. Mortamais M, Reynes C, Brickman AM, et al. Spatial distribution of cerebral white matter lesions predicts progression to mild cognitive impairment and dementia. PLoS One 2013; 8(2):e56972. 32. Okonkwo OC, Alosco ML, Jerskey BA, et al. Cerebral atrophy, apolipoprotein E varepsilon4, and rate of decline in everyday function among patients with amnestic mild cognitive impairment. Alzheimers Dement 2010; 6(5): 404–411. 33. Misra C, Fan Y, Davatzikos C. Baseline and longitudinal patterns of brain atrophy in MCI patients, and their use in prediction of short-term conversion to AD: results from ADNI. Neuroimage 2009; 44(4):1415–1422. 34. Leow AD, Yanovsky I, Parikshak N, et al. Alzheimer’s disease neuroimaging initiative: a one-year follow up study using tensor-based morphometry correlating degenerative rates, biomarkers and cognition. Neuroimage 2009; 45(3):645–655. 35. Tapiola T, Pennanen C, Tapiola M, et al. MRI of hippocampus and entorhinal cortex in mild cognitive impairment: a follow-up study. Neurobiol Aging 2008; 29(1):31–38. 36. Eckerstrom C, Andreasson U, Olsson E, et al. Combination of hippocampal volume and cerebrospinal fluid biomarkers improves predictive value in mild cognitive impairment. Dement Geriatr Cogn Disord 2010; 29(4): 294–300. 37. Mueller SG, Schuff N, Yaffe K, et al. Hippocampal atrophy patterns in mild cognitive impairment and Alzheimer’s disease. Hum Brain Mapp 2010; 31(9):1339–1347. 38. Gao FQ, Black SE, Leibovitch FS, et al. Linear width of the medial temporal lobe can discriminate Alzheimer’s disease from normal aging: the Sunnybrook dementia study. Neurobiol Aging 2004; 25(4):441–448. 39. Uotani C, Sugimori K, Kobayashi K. Association of minimal thickness of the medial temporal lobe with hippocampal volume, maximal and minimal hippocampal length: volumetric approach with horizontal magnetic resonance imaging scans for evaluation of a diagnostic marker for neuroimaging of Alzheimer’s disease. Psychiatry Clin Neurosci 2006; 60(3):319–326. 40. Gabryelewicz T, Pawlowska-Detko A, Misko J, et al. Prediction of deterioration of mild cognitive impairment with CT and SPECT. Med Sci Monit 2007; 13(Suppl 1):31–37. 41. Prins ND, van Dijk EJ, den Heijer T, et al. Cerebral white matter lesions and the risk of dementia. Arch Neurol 2004; 61(10):1531–1534. 42. Boutet C, Chupin M, Colliot O, et al. Is radiological evaluation as good as computer-based volumetry to assess hippocampal atrophy in Alzheimer’s disease? Neuroradiology 2012; 54(12):1321–1330. 43. Eckerstrom C, Olsson E, Bjerke M, et al. A combination of neuropsychological, neuroimaging, and cerebrospinal fluid markers predicts conversion from mild cognitive impairment to dementia. J Alzheimers Dis 2013; 36(3):421–431.