Marcel Kettelmann1, Stephan Niland2, Mirjam Gerwing1, Markus Wick3, Sascha Koehler3, Lydia Wachsmuth1, Moritz Wildgruber1, Johannes A Eble2, and Cornelius Faber1
1Translational Research Imaging Center (TRIC), University Hospital Muenster, Muenster, Germany, 2Institute of Physiological Chemistry and Pathobiochemistry, University Hospital Muenster, Muenster, Germany, 3Bruker BioSpin, Ettlingen, Germany
Synopsis
Diffusion
weighted MRI using oscillating gradients has previously been shown to provide
microstructural information on celllularity in tumors, which may serve as
marker for monitoring therapy response. Which geometrical model for analysis of
such data provides most reliable results, is, however, a matter of debate.
Here, we used the IMPULSED approach, and show that cell radii in tumor
spheroids of three different cell lines can be determined with high accuracy.
In an in vivo model, however, deviations from radii determined by laser
scanning microscopy were found.
Intrduction
DW-MRI has been shown to provide a sensitive early marker for therapy response in
tumor treatment. Changes in tissue density, cellularity, or vascularization are
reflected in diffusion parameters. How exactly structural and physiological
parameters influence diffusion parameters, is currently a matter of debate. MRI
sequences using oscillating instead of constant pulsed gradients, have been
shown to provide more detailed insight into the microstructure of tumors. Three
different geometrical models dubbed IMPULSED, POMACE and VERDICT have
previously been described, as reviewed in (1). Here, we have used oscillating
gradient DWI and the IMPULSED approach to calculate diameters of cells in
spheroids of three different cancer cell lines, and in vivo in human xenograft
mouse model, and we compared those to laser scanning microscopy data.Methods
Tumor spheroids of SK-Mel-30, MG63 and A375 cells,
respectively, were grown in 96 well plates to a diameter of 500-700 µm. For
this purpose, 3000 or 6000 cells were initially seeded in methylcellulose
containing medium, grown overnight and stained with wheat germ agglutinin, tetramethylrhodamine
conjugate (WGA-TRITC) and Hoechst 33342. Spheroids were transferred to PCR
tubes and embedded in Iow-melting agarose for MRI (Fig. 1). For the in vivo
model 106 4T1 cells were injected into the mammary fat pad of balb-c
mice and grown for three to six days (n=3). Scanning was performed at 9.4 T on
a 94/20 BioSpec (Bruker BioSpin) using a CryoProbe and a 0.7 mT/m gradient
system. Images were acquired with a spin echo gradient sequence using
oscillating gradients with the following parameter for spheroids:
(TR/TE:2000/89.4 ms; 2 segments, matrix 128x128, FOV: 8x8 mm2,
resulting in a resolution of 63 µm and a scan time of 1:59 min). Modulation frequencies
of the 20 ms gradients (5 ms separation) were 200 Hz, 150 Hz, 100 Hz and 50 Hz.
For each frequency nine b-values were used: 1000, 875, 750, 625, 500, 375, 250,
125, 0 s/mm2. Four repetitions were performed for each set of scan
parameters. Parameters for in vivo scans were TR/TE:3000/21.5 ms; 2 segments,
matrix 108x96, FOV: 18x15 mm2)
For
analysis according to the IMPULSED approach, perfectly spherical cells with
impermeable, infinitesimal walls were assumed. Tumors were manually segmented
in the images and intensities fitted with the equation given by Jiang et al (2,3),
yielding five parameters including average cell radius.
Laser scanning microscopy was performed with a LSM
800 (Zeiss) (Fig. 2). Cell radii and eccentricity were determined manually
using ImageJ.Results
SNR values ranged between 30 and 70, and allowed for
clearly delineating the spheroids and successful fitting of the data. For
A375, SK-Mel-30, and MG-63 cells average radii of 11.1±0.9 µm (n=6 spheroids), 11.7±1.3
µm (n=9), and 13.9±1.4 µm (n=7) were calculated from the DWI data. These agreed
well with the values obtained for the different cell lines by laser scanning
microscopy of 8.7±2.4 µm (n=41 cells), 11.5±3.0 µm (n=91), and 13.6±2.1 µm
(n=50). Eccentricity values were determined as 0.34, 0.42, and 0.34,
respectively. Radii obtained by MRI for each individual spheroid and average
values from microscopy are summarized in Fig. 3.
For in vivo experiments, MR analysis of the 4T1
cells yielded an average radius of 9.9±1.9 µm, which exceeded the radius
obtained from laser scanning microscopy of 6.4±1.0 µm (Fig. 4). The eccentricity
was determined as 0.72.Discussion
Determining
cell radii from oscillating gradient DW-MRI is based on assumptions about
geometry and exchange between compartments. Tumor spheroids very closely
reproduce the assumption of spherical cells with no other disturbing
structures, such as vessels or necrotic tumor regions. For these conditions the
IMPULSED approach is capable to determine cell radii with high accuracy, as shown
here for three different tumor cell lines. For tumors in vivo, MRI results
deviated from microscopy data. This mismatch may be explained in part by the
larger eccentricity of the 4T1 cells, but is most likely also due to the
presence of vessels and small areas with beginning necrosis.Conclusion
The IMPULSED approach provides highly accurate radii
for spherical cells in tumor spheroids. For in vivo tumors however, more
complex geometrical assumptions, such as for example the VERDICT model (4), may
be required to provide correct estimates of the cellular microstructure, a
potentially reliable marker for treatment response in cancer.Acknowledgements
This work was supported by the DFG SFB1009, Z02 and A09
References
1. Reynaud O.
Time-dependent diffusion MRI in cancer: Tissue modeling and applications. Front Phys. 2017; 5:1-16.
2. Jiang X, Li H, Xie J, et al. Quantification of
cell size using temporal diffusion spectroscopy. Magn Reson Med. 2016; 75:1076-1085.
3. Jiang X, Li H, Xie J, et al. In vivo imaging of
cancer cell size and cellularity using temporal diffusion spectroscopy. Magn Reson Med.
2017; 78:156-164.
4. Panagiotaki
E, Walker-Samuel S, Siow B, et al. Noninvasive quantification of solid tumor
microstructure using VERDICT MRI. Cancer Res.
2014; 74:1902–12.