Table 1 Summary of combined magnetic resonance imaging and cerebrospinal fluid studies in Alzheimer's disease

Schönknecht et al., 2003 [57]

88 AD, 17 CN

In AD, CSF tau was not correlated to MRI.

Wahlund and Blennow, 2003 [48]

23 MCI, 24 AD

At baseline, CSF Aβ_1-42 was correlated with MRI. During the follow-up period, increases in tau and p-tau correlated with ventricular increase.

de Leon et al., 2004 [46]

32 stable CN, 13 CN progressed to MCI

Accuracy for prediction of CN progression to MCI: Baseline: MRI: 78%; CSF: 78% to 89%.

Hippocampal volume decrease correlates with P-tau₂₃₁ increase and Aβ_1-42 decrease.

Hampel et al., 2005 [73]

22 AD

CSF p-tau₂₃₁ correlated with baseline hippocampus and rates of hippocampal atrophy.

Schoonenboom et al., 2005 [74]

39 MCI

CSF Aβ_1-42 was correlated with MRI and not tau.

de Leon et al., 2006 [42]

9 CN, 7 MCI

Accuracy for separation of CN and MCI: Baseline: MRI: 94%, CSF: 63% to 88%; MRI + CSF: 94% Longitudinal: MRI: 88%; CSF: 73% to 88%; MRI + CSF: 94%

In MCI, longitudinal hippocampal volume decrease correlated with P-tau₂₃₁ increase and Aβ_1-42 decrease.

Herukka et al., 2008 [51]

21 MCI, of whom 8 progressed to AD

In all MCI, increases in tau and p-tau correlated with a decrease in hippocampal volumes.

Schoonenboom et al., 2008 [56]

32 CN, 61 AD

Odds ratio between AD and CN: MRI: 28; CSF: 57

There were no correlations between visual assessment of MRI and CSF biomarkers within CN and AD.

Sluimer et al., 2008 [52]

23 CN, 9 MCI, 47 AD

In AD, CSF p-tau₁₈₁ had mild association with whole-brain atrophy rate. Only MRI was associated with change in cognitive measures.

Brys et al., 2009 [45]

21 CN, 16 stable MCI, 8 MCI progressed to AD

Accuracy for prediction of MCI progression to AD: MRI: 74%; CSF: 70%; MRI + CSF: 84%

There were no longitudinal correlations between MRI and CSF.

Chou et al., 2009 [53]

80 CN, 80 MCI, 80 AD (ADNI)

CSF Aβ_1-42 was correlated with ventricular expansion.

Fagan et al., 2009 [54]

69 CN, 29 mild AD

In CN, decrease in CSF Aβ_1-42 correlated with brain atrophy. In mild AD, increases in CSF t-tau and p-tau₁₈₁ correlated with brain atrophy.

Henneman et al., 2009 [49]

19 CN, 25 MCI, 31 AD

Baseline CSF p-tau₁₈₁ was independently associated with subsequent disease progression, measured by hippocampal atrophy rate.

Leow et al., 2009 [75]

40 CN, 40 MCI, 20 AD (ADNI)

Baseline CSF correlated with temporal atrophy rates over the course of 12 months.

Schuff et al., 2009 [76]

112 CN, 226 MCI, 96 AD (ADNI)

In MCI, an increase in rates of hippocampal atrophy correlated with lower CSF Aβ_1-42.

Thomann et al., 2009 [50]

15 CN, 23 MCI (AACD), 16 AD

Increases in CSF t-tau and p-tau₁₈₁ correlated with cortical atrophy in temporal, parietal, and frontal regions.

Vemuri et al., 2009 [43]

109 CN, 192 aMCI, 98 AD (ADNI)

AUROC separating CN, aMCI, and AD: MRI: 0.77; CSF: 0.68 to 0.73; MRI + CSF: 0.81

Within each clinical group, only MRI correlated with cognition in aMCI and AD groups.

Vemuri et al., 2009 [47]

109 CN, 192 aMCI, 98 AD (ADNI)

Proportional hazards for predicting time to conversion from aMCI to AD: MRI: 2.6; CSF: 1.7 to 2.0

Baseline MRI was a better predictor of subsequent cognitive and functional decline than baseline CSF was.

Vemuri et al., 2010 [34]

92 CN, 149 MCI, 71 AD (ADNI)

Sample size required to detect treatment effects in AD: MRI: 100; CSF >10⁵.

Longitudinal annual changes were observed only in MRI and not in CSF. Change in MRI was associated with change in cognitive measures.

Walhovd et al., 2010 [44]

42 CN, 73 MCI, 38 AD (ADNI)

Accuracy for baseline separation of CN and AD: MRI: 85%; CSF: 81.2%; CSF + MRI: 88.8%

In MCIs, only baseline MRI and FDG were correlated to (or predictive of) future clinical decline during 2 years.

Fjell et al., 2010 [77]

71 CN

Below a certain threshold, baseline CSF Aβ_1-42 correlated with ventricular increase and volumetric brain decrease over the course of 1 year.

Fjell et al., 2010 [55]

Baseline: 105 CN, 175 MCI, 90 AD (ADNI)

In MCI and AD, baseline CSF measures were not related to baseline MRI but were related to longitudinal atrophy. Baseline MRI predicted change in cognition better than CSF did.