Eur Radiol (2009) 19: 898–903 DOI 10.1007/s00330-008-1210-8
Agata Majos Piotr Bogorodzki Ewa Piątkowska-Janko Tomasz Wolak Robert Kurjata Ludomir Stefańczyk
Received: 12 April 2008 Revised: 20 August 2008 Accepted: 23 August 2008 Published online: 6 November 2008 # European Society of Radiology 2008 A. Majos (*) . L. Stefańczyk Radiology Department, Medical University of Lodz, Kopcinskiego 22, 90-153 Lodz, Poland e-mail:
[email protected] P. Bogorodzki . E. Piątkowska-Janko . R. Kurjata Institute of Radioelectronics, Warsaw University of Technology, Nowowiejska 15/19, 00-665 Warsaw, Poland T. Wolak Institute of Physiology and Pathology of Hearing, Pstrowskiego 1, 01-943 Warsaw, Poland
CONTRAST MEDIA
Functional imaging with MR T1 contrast: a feasibility study with blood-pool contrast agent
Abstract The aim of this study was to prove the concept of using a long intravenous half-life blood-pool T1 contrast agent as a new functional imaging method. For each of ten healthy subjects, two dynamic magnetic resonance (MR) protocols were carried out: (1) a reference run with a typical T2* echo-planar imaging (EPI) sequence based on the blood oxygenation level-dependent (BOLD) effect and (2) a run with a T1-sensitive threedimensional (3D) gradient-echo (GRE) sequence using cerebral blood volume (CBV) contrast after intravenous administration of a contrast agent containing a chelate of gadolinium diethylene-triamine-pentaacetate with a phosphono-oxymethyl substituent. All sequences were performed during the execution of a block-type fingertapping paradigm. SPM5 software
Introduction Over last decade, functional magnetic resonance imaging (fMRI) has evolved from an innovative laboratory technique into a widely used method for the investigation of brain functional anatomy in neurology, psychiatry and neurosurgery. A majority of fMRI experiments are based on the blood oxygenation level-dependent (BOLD) effect. Although the neuronal activity is not measured directly, it leads to the complex, local haemodynamic responses in cerebral blood flow (CBF), cerebral blood volume (CBV), and cerebral metabolic rate of oxygen (CMRO2). These processes result in changes in the balance between two
was used for statistical analysis. For both runs maximum activations (peak Z-score=5.5, cluster size 3,449 voxels) were localized in the left postcentral gyrus. Visual inspection of respective signal amplitudes suggests the T1 contrast to be substantially smaller than EPI (0.5% vs 1%). A new functional imaging method with potentially smaller image artefacts due to the nature of CBV contrast and characteristics of the T1 sequence was proposed and verified. Keywords Blood oxygen level-dependent effect . CBV . Functional MRI . Contrast media
forms of haemoglobin, which possess different magnetic properties: oxyhaemoglobin (diamagnetic) and deoxyhaemoglobin (paramagnetic), detected by the most frequently used gradient-echo (GRE) sequences like the single-shot echo-planar imaging (SSEPI) sequence [1–4]. CBV functional maps, which were the first measurements of cortical activity of the brain, can serve as a complement to BOLD imaging. The introduction of paramagnetic contrast agents (MION and USPIO) increased their value for monitoring the temporal and spatial brain activity, which has been demonstrated already [5, 6]. CVB maps serve, in general, to improve spatial specificity of fMRI experiments because they eliminate signal from
899
large veins and thus limit functional response primarily to capillaries [7, 8]. Dilatation of regional capillary beds as a response to neuronal activation results in increased amounts of contrast medium (CM) in specific nervous centres. That provides stronger changes in MR signal due to the high magnetic susceptibility of CM. Additionally, the physical properties of MR sequences have a significant impact on the last result in creating functional maps. SSEPI in known to suffer from signal drop-out near susceptibility boundaries, resulting in distortion and blurring on produced images [9]. This tremendously limits clinical application of functional examinations in cases when regions of interest border with air, dense bone structures, i.e. paranasal sinuses, trephine cavities, temporal bones or surgical implants. These limitations could potentially be overcome by using a T1-weighted sequence, together with the administration of a blood-pool contrast agent that can remain in the vascular bed long enough to allow for functional imaging. Early imaging studies involving animals confirm the feasibility of such a procedure [10].
Materials and methods Ten healthy right-handed participants (four males and six females, age 21–52 years) were selected from a group of patients undergoing MRI examinations with administration of CM for diagnosis of peripheral vascular disease. Eligibility criteria for the volunteers consisted of the absence of any pre-existing or present abnormal neurological and vascular conditions. All subjects were fully informed as to the nature of the study and all gave their consent regarding the participation. The local ethics committee at the Medical University of Lodz approved the procedures utilized in the study, which was conducted in accordance with the principles set by the Helsinki Convention. For each of them the two functional ‘runs’ were acquired on a 1.5-T MR system (Siemens, Avanto, Erlangen, Germany). First, a high-resolution T1 anatomical MR sequence was performed with a 1×1×1-mm isotropic voxel. Following that, a reference EPI (EPI) was performed with a typical T2* sensitive GRE EPI sequence (TR=3 s/ TE=50 ms/FA=90 degrees) with 29 3-mm-thick slices with 0.75-mm gap and 3.94-mm in-plane resolution. At 15–20 min after intravenous administration (0.03 mmol/ kg) of Vasovist (Bayer Schering, Berlin), a second T1-
left
Fig. 1 Alternating left- and right-hand finger-tapping time-block periods 0
right
50
left
sensitive (T1) sequence was performed, with a 3D GRE sequence (TR=8.5 ms/TE=3.14 ms/FA=10 degrees) with 16 5-mm-thick slices and 0.78-mm in-plane resolution [11]. Vasovist, previously known as MS-325, is a contrast agent containing a chelate of gadolinium diethylenetriamine-pentaacetate with a (diphenylcyclohexyl)phosphono-oxymethyl substituent (gadofosveset trisodium) at a concentration of 0.25 mmol/ml. Following intravenous administration, it reversibly binds to the albumin plasma protein (80–87%), thus extending plasma half-life (0.48± 0.1 h for t½α, and ~16 h for t½β) and limiting its movement out of blood vessels. The relaxation value of the plasma at 20 MHz is calculated between 33.4 and 45.7 mM−1s−1 (for doses of 0.05 mmol/kg mc) [12–14]. Both these characteristics delineate the value of Vasovist for fMRI with T1dependent sequences [15]. Vasovist was administered in doses recommended by the manufacturer for standard examinations: MR angiography of abdominal and lower extremity vessels, i.e. 0.12 ml/kg mc, which is equal to 0.03 mmol/kg mc. During both dynamic MR protocols (EPI and T1) subjects were asked to perform a block-type paradigm with 15 finger tapings in 30-s length ‘on’ periods, followed by identical length ‘off’ resting periods (Fig. 1). Five repetitions of the above method took 300 s, giving 100 volumes acquired in EPI, and 30 volumes in T1. Data were transferred from DICOM to ANALYZE NIFTI format using the SPM5 [Functional Imaging Laboratory, www.fil. ion.ucl.ac.uk/spm/software/spm5/, Matlab (Mathworks, USA)] toolbox for further analysis steps. Image preprocessing and statistical analyses were conducted in SPM5. First, an intra-run motion correction was accomplished with the first image as a reference. Realigned data were time corrected in order to minimize errors caused by volume-slice MR acquisition times. After realignment and time correction, data were normalized to stereotaxic space with an isotropic 2×2×2-mm voxel size. Gaussian kernel [full width half maximum (FWHM)=6 mm] smoothing was applied thereafter in order to minimize spatial interpolation errors. Statistical parametric maps (SPM) were generated using the general linear model (GLM) with a haemodynamically corrected [canonical haemodynamic response (HRF)] box-car function as the reference paradigm. Furthermore, given that we had an a priori hypothesis regarding the activity within the Brodmann area (BA) 2 (primary somatosensory cortex), we created a region of interest (ROI) mask using the Anatomy toolbox utility to
right
100
left
right
150
Time in seconds
left
200
right
left
250
right
300
900
Table 1 RFX one-sample T-test for EPI and T1 runs
Run
Contrast
Peak Z-score
Cluster size
MNI max location
% of cluster size located in BA
EPI
Left
4.7
679
34 −22 62
Right
5.6
949
−32 −34 62
Left
4.4
271
38 −28 42
Right
4.3
51
−40 −30 44
BA2(R) - 1 BA3(R) - 14 BA4(R) - 21 BA6(R) - 5 BA40(R) - 2 BA1(L) - 1 BA2(L) - 4 BA3(L) - 15 BA4(L) - 18 BA6(L) - 1 BA 40(L) - 7 BA2(R) - 9 BA3(R) - 24 BA4(R) - 21 BA2(L) - 16 BA3(L) - 14 BA4(L) - 10 BA40(L) - 25
T1
restrict analyses to this region only. This ROI was used for each of the subjects in order to calculate a percent signal change (PSG) measure [16, 17]. A “second level” (SL) statistical test were performed thereafter on SPMs in order to evaluate group specific effects. Group contrast maps were created for the two runs to determine the mean suprathreshold activation for each run. This difference in activation was displayed as SPM{t} maps over an average template brain in the standardized coordinate space of the Montreal Neurological Institute (MNI). Using a random effects (RFX) analysis, we made direct comparisons between data from the both functional runs.
Percent signal changes in the primary somatosensory cortex for finger-tapping-induced activations in EPI and T1 experiments are presented in Fig. 4.
Discussion In order to detect the BOLD effect, T2*-weighted sequences are commonly used. These sequences are characterized by high sensitivity to non-uniformities in the magnetic field. This enables the acquisition of a good
Results An RFX one-sample T-test for EPI and T1 runs is presented in Table 1 and Fig. 2. Data were thresholded on T>5 (p< 10e-4 corrected on cluster-level) and cluster size k>16. For both runs maximum activations (peak Z-score=5.5, cluster size 3449 voxels) were localized in the left postcentral gyrus spreading over BA1, BA2, BA3, BA4, BA6 and BA40. Example signal intensity time-courses were extracted from one of the subjects’ runs, placing a 10-mm diameter sphere around the location of peak activation either for two runs (EPI, T1) or contrasts (left, right). Visual inspection of respective signal amplitudes suggests the T1 contrast to be substantially smaller than EPI (0.5% vs 1%) (Fig. 3)
Fig. 2 Results of the group level T-test (random effect analysis) for two contrasts (left hand finger tapping in blue, and right in red) for EPI and T1 runs
901
LH RH
2.5
EPI
2 1.5
MR signal in %
Fig. 3 Example time courses of MRI signal changes (divided by the base line and scaled in percents) obtained from 10 mm radius sphere located in left (red) and right (blue) motor areas in one of the subjects (AM). Dots correspond to scanning instants (3 s in EPI run and 10 s in T1 run) LH left hand, rh right hand
1 0.5 0 -0.5 -1 -1.5 0
50
100
150
200
250
300
Time in seconds 1
LH RH
T1
0.8 0.6
MR signal in %
0.4 0.2 0 -0.2 -0.4 -0.6 -0.8 -1 0
50
100
150
200
250
300
Time in seconds
BOLD signal but is also a source of serious artefacts. Differences in magnetic susceptibility of air, bone, different types of tissues, as well as surgical implants create geometric artefacts, which ultimately decrease image quality [9, 10]. Therefore, T2*-weighted sequences do not allow for the precise functional evaluation of brain regions that border with these anatomical structures, for example, at the base of the frontal lobes, near the collateral sinuses, the temporal lobes, which are adjacent to the petrous pyramids, or the mastoid cells, or post-trepanation cavities [18, 19]. This limitation can have significant clinical implications when it involves brain regions where important nervous centres are located. There are some approaches reducing the above drawbacks: combinations of spin and GRE techniques,
shimming parts of the brain that are outside of the area of interest [10]. However, these methods are characterized by markedly reduced sensitivity in comparison with the T2* sequence, because of the loss of static spin dephasing. A new and promising solution to these limitations would be to use a T1-weighted sequence with the administration of a blood-pool contrast agent, a procedure already demonstrated in early experimental work [20]. Animal studies have proved that long intravenous half-life contrast agents, monocrystalline iron oxide nanoparticles (MION) and gadolinium-based MS-325, can be successfully used in fMRI [8, 11]. Neuronal activation causes a local increase in blood volume via arteriole and capillary dilation. Consequently,
902
Fig. 4 Percent signal changes in the primary somatosensory cortex (Area 2, in SPM Anatomy toolbox 1.5) in EPI and T1 experiments
the amount of the contrast agent increases in the stimulated regions of brain. Thanks to the long duration of CM binding with albumin, the relaxation value of the elicited tissue is significantly shortened, which induces MR signal in the T1 sequence efficiently. The important fact is that the relaxation values of gadofosveset have been shown to last stably for 4 h after its intravenous administration. This is sufficient time to carry out a typical fMRI experimental procedure. Analysis of data between the two groups conducted for our study revealed that while performing the same motor task, the T2* BOLD and T1 Vasovist activated areas overlapped. A statistically significant activation was found in all the participants using both the T2* BOLD and the T1 Vasovist techniques. The activations obtained for the T1 Vasovist group were characterized by a smaller volume of activation and a smaller percent change in the signal.
However, the increase of signal intensity in T1 Vasovist sequences depends on changes in CBV only in opposition to the BOLD effect, which relies also on CBF and CMRO2 changes. It could be one of the main reasons for divergence in the observed results between these two dynamic techniques. At the same time, CBV contrast does not suffer from artefacts related to blood flow, which appears to be an important advantage of this method, because allows for a more precise localization of the activation [5, 7]. None of our examinations revealed activated voxels in the ventricles or outside the skull bones, something that is common in classic EPI BOLD imaging. Another reason for certain discrepancies in results between BOLD and CBV runs is probably lack of optimization of parameters for T1-weighted sequence. We used one of the routine protocols in our study but believe that modifications provided for functional examinations allow for much better results.
903
A limitation of the T1 Vasovist technique is that it is an invasive procedure, on account of the administration of a contrast agent. However, this method can undoubtedly be employed in situations where the contrast agent is administered mandatorily due to the routine clinical indications or in specific situations requiring the evaluation of centres in specific anatomical areas. Then it could become an additional source of relevant information. In the present study, Vasovist was administered in doses recommended by the manufacturer. Further work is needed to determine whether it is necessary to optimize the doze in order to improve the signal acquisition in functional imaging.
Conclusions A new functional imaging method with potentially smaller image artefacts, due to the nature of CBV contrast and characteristics of T1 sequence, was proposed and verified. The technique of long intravenous blood-pool agents, together with high spatial resolution T1 sequence could be a promising perspective for fMRI examinations in nervous centres sensitive to susceptibility artefacts.
References 1. Nair DG (2005) About being BOLD. Brain Res Brain Res Rev 50:229–243 2. Buxton RB (2001) Introduction to functional magnetic resonance imaging: principles and techniques. Cambridge University Press, New York 3. Arthurs OJ, Boniface S (2002) How well do we understand the neural origins of the fMRI BOLD signal? Trends Neurosci 25:27–31 4. Iannetti GD, Wise RG (2007) BOLD functional MRI in disease and pharmacological studies: room for improvement? Magn Reson Imaging 25:978–988 5. Mandeville B, Marota J, Kosovsky B, Keltner JR, Weissleder R, Rosen B, Weisskoff R (1998) Dynamic functional imaging of relative cerebral blood volume during rat forepaw stimulation. Magn Reson Med 39:615–624 6. Reese T, Bjelke B, Porszasz R, Baumann D, Bochelen D, Sauter A, Rudin M (2000) Regional brain activation by bicuculline visualized by functional magnetic resonance imaging. Time-resolved assessment of bicuculline-induced changes in local cerebral blood volume using an intravascular contrast agent. NMR Biomed 13:43–49 7. Smirnakis SM, Schmid MC, Weber B, Tolias AS, Augath M, Logothetis NK (2007) Spatial specificity of BOLD versus cerebral blood volume fMRI for mapping cortical organization. Cereb Blood Flow Metab 27:1248–1261
8. Keilholz SD, Silva AC, Raman M, Merkle H, Koretsky AP (2006) BOLD and CBV-weighted functional magnetic resonance imaging of the rat somatosensory system. Magn Reson Med 55:316–324 9. Frahm J, Merboldt KD, Hänicke W (1988) Direct FLASH MR imaging of magnetic field inhomogeneities by gradient compensation. Magn Reson Med 6:474–480 10. Morton DW, Keogh B, Lim K, Maravilla KR (2006) Functional brain imaging using a long intravenous halflife gadolinium-based contrast agent. AJNR Am J Neuroradiol 27:1467– 1471 11. Buxton RB, Edelman RR, Rosen BR, Wismer GL, Brady TJ (1987) Contrast in rapid MR imaging: T1- and T2weighted imaging. J Comput Assist Tomogr 11:7–16 12. Parmelee DJ, Walovitch RC, Ouellet HS, Lauffer RB (1997) Preclinical evaluation of the pharmacokinetics, biodistribution, and elimination of MS325, a blood pool agent for magnetic resonance imaging. Invest Radiol 32:741–747 13. Rohrer M, Bauer H, Mintorovitch J, Requardt M, Weinmann HJ (2005) Comparison of magnetic properties of MRI contrast media solutions at different magnetic field strengths. Invest Radiol 40:715–724 14. Giovagnoni A, Catalano C (2007) Application of blood-pool agents in visualization of peripheral vessels. Eur Radiol 17(Suppl 2):B18–B23
15. Caravan P, Parigi G, Chasse JM, Cloutier NJ, Ellison JJ, Lauffer RB, Luchinat C, McDermid SA, Spiller M, McMurry TJ (2007) Albumin binding, relaxivity, and water exchange kinetics of the diastereoisomers of MS–325, a gadolinium(III)-based magnetic resonance angiography contrast agent. Inorg Chem 46:6632–6639 16. Grefkes C, Geyer S, Schormann T, Roland P, Zilles K (2001) Human somatosensory area 2: observerindependent cytoarchitectonic mapping, interindividual variability, and population map. Neuroimage 14:617–631 17. Geyer S, Ledberg A, Schleicher A, Kinomura S, Schormann T, Bürgel U, Klingberg T, Larsson J, Zilles K, Roland PE (1996) Two different areas within the primary motor cortex of man. Nature 382:805–807 18. Howseman AM, Josephs O, Rees G, Friston KJ (1997) Special issues in functional resonance imaging. In: Frackowiak RSJ et al (ed) Human brain function. Academic Press, San Diego, pp 1–20 19. Truong TK, Clymer BD, Chakeres DW, Schmalbrock P (2002) Threedimensional numerical simulations of susceptibility-induced magnetic field inhomogeneities in the human head. Magn Reson Imaging 20:759–770 20. Jochimsen TH, Norris DG, Mildner T, Möller HE (2004) Quantifying the intra- and extravascular contributions to spin-echo fMRI at 3 T. Magn Reson Med 52:724–732
