Synopsis
There continues to be interest in using changes in CSF properties to image neurodegenerative diseases. To optimize MRI sequences that enable segmentation of CSF from tissue, we characterized the CSF relaxometric properties at various field strengths in vivo and in vitro. Our in vitro results suggest that in vivo T2 value at high field is incorrect due to residual gradients and that low field is more optimal to quantify CSF relaxivity in vivo. We have shown an important difference of in vitro CSF T2 vs saline T2 that is mostly explained by the relaxivity of glucose.Purpose
There continues to be interest in using changes in CSF properties to image neurodegenerative diseases
1,2. Due to the different relaxation properties of CSF and tissue there are many MRI sequences that enable segmentation of CSF from tissue
3-4. Optimization of these sequences requires understanding CSF relaxometric properties. While T
2 values in the brain tissue have been published, there is a poor literature about CSF relaxation times. In this study, we aimed to establish the values of T
2 for
in vivo human and mice CSF at various field strengths and for in vitro CSF samples at 14.1T. We aimed also to determine whether the T
2 values are affected by physiological protein, metals and glucose CSF concentration.
Methods
IN VIVO Human: 6 healthy patients were scanned (3 per field). MRI: T2 map were performed on a 3 and 7T MRI system. Imaging was performed using a spin echo multi-contrast sequence with TR=10000ms, TE(32)=30.5-945.5 ms and a resolution of 3 mm3. Animals: 12 healthy C57Bl6 mice were used (3 per field). All procedures were performed under isoflurane anesthesia (2.5%). MRI: T2 was calculated on a 1, 4.7, 9.4 and 11.7T MRI system. Spectroscopy was performed using a mutiecho CPMG sequence with TR=20000ms, TE(48)=1.875–1029ms. Post-Processing: T2 map was calculated by performing a mono-exponential fit of the pixel or echo intensities. For human, ROIs were drawn on the lateral ventricles and T2 map was fitted using the least absolute residuals method. For mice, T2 map was fitted using a non-negative least squares method5 and spectrums were analyzed.
IN VITRO Human: CSF of 3 healthy patients was obtained from another protocol via a lumbar puncture. Animals: CSF was removed via a transcutaneous cisterna magna puncture in 3 healthy rhesus macaque monkey and 4 CD rats. CSF protein quantification: was performed according to the Bradford method6. CSF metals quantification: Cu, Fe, Mn and Zn were quantified by inductively coupled plasma mass spectrometry (Exova, California). Solutes preparation: all solutes were diluted in saline: bovine serum albumin (BSA, 10-200 mg/dL), ZnCl2, FeCl2, CuCl2 (0-1 mM), MnCl2 (0-80 µM), haptoglobulin (0-0.1 mg/dL), α2-macroglobulin (0-1.2 mg/dL), human transferrin (0-5 mg/dL) and glucose (0-1000 mg/dL). MRI: T2 values were calculated at 37°C on a 14.1 T MRI system. T2 Measurement: see above (animals section). Post-Processing: see above (animals section). For each solute, ΔR2 (1/T2) values were plotted against solute concentration and relaxivity r2 was computed (ΔR2 = r2 x [solute] ; Figure 1). For each solute, concentration in human healthy CSF was calculated (BSA, metals) or estimated from literature (transferrin7, haptoglobulin8, α2-macroglobulin9, glucose10). Based on this concentration, ΔR2 value in healthy human CSF for each solute was extrapolated. Finaly, we calculated the percentage of contribution of each solute to change human healthy CSF T2.
Results
In vivo, CSF T
2 decrease with increase in field strength for both human and rodent (Table 1).
In vitro, the differences in T
2 between saline and CSF were large (~ 40%; Table 2).
In vitro, CSF T
2 was comprised between 1.4-2 s corresponding to the in vivo CSF T
2 values at low field. To test whether paramagnetic metals ions, proteins and/or glucose could explain the T
2 relaxivity of CSF, we calculated they concentration (Figure 1). For all metals, the concentration was too low to modify significantly the CSF T
2 relaxivity (<0.5%). For all proteins, there are able to change 13% of the CSF T
2. Finally, the glucose concentration is high enough to contribute to a CSF T
2 change of 54%.
Discussion
T
2 in vivo measurements at low field (1T) are in agreement with literature
11. Hopkins and al., have shown a T
2 between 1.8-2.2 s measured in humans
in vivo at a very low field
11. That is also in agreement with our T
2 in vitro measurement
11-12-13. However, our results suggest that
in vivo T
2 value at high field is incorrect. This is likely due to a: residual applied gradient, residual field gradient from tissue and CSF and a gradient shimming. Therefore, low field is more optimal to quantify CSF relaxivity
in vivo. In this study we were looking for the compound that is able to significantly change CSF T
2 relaxivity. Our results suggest that concentration of metals in CSF is too low to significantly change the CSF T
2. CSF T
2 was also not significantly affected by protein concentrations. Interestingly, we discovered that glucose is significantly able to change CSF T
2. We will confirm this result with ongoing
in vivo studies. These data are important for the development of new MRI sequence for CSF segmentation and for possible detection of glucose variation within the CSF.
Acknowledgements
This study was supported by the Intramural Research Program of the National Institute of Neurologic Disorders and Stroke (NINDS), National Institutes of Health.References
1Abdullah et al., Conf Proc IEEE Eng Med Biol Soc, 2013, 2013:3359-62; 2Gao et al., Neuroimage, 2014, 100:370-8; 3Hodel et al., Eur Radiol, 2013, 24(1):136-42; 4Poh et al., Neuroinformatics, 2012, 10(2):159-72.; 5Prasloski et al., Magn Reson Med, 2012, 67(6):1803-14; 6Bradford, Anal Biochem, 1976, 72:248-54; 7LeVine et al, Brain Research 1999, 821: 511–515 ; 8Chuang et al., Clinica Chimica Acta , 2013, 417: 26–30 ; 9Kanoh et al., Bioch. and mol. Biol. Inter., 1997, 43 : 269-278 ; 10Karen Roos (2005), Medical Pub. Division. p. 4. ISBN 0-07-140816-9; 11Hopkins et al., Magn Reson Med, 1986, 3(2):303-11; 12Larsson et al., Magn Reson Med, 1989, 11(3):337–348 ; 13Jezzard et al., Radiology, 1996, 199(3):773-9.