Download Microscopic diffusion anisotropy in formalin fixed prostate tissue

NOTE
Magnetic Resonance in Medicine 000:000–000 (2012)
Microscopic Diffusion Anisotropy in Formalin Fixed
Prostate Tissue: Preliminary Findings
Roger M. Bourne,1* Nyoman Kurniawan,2 Gary Cowin,2
Paul Sved,3 and Geoffrey Watson1
Diffusion tensor microimaging at 16.4 T with 40 mm isotropic
voxels was used to investigate anisotropic water diffusion in
prostate tissue at spatial resolution approaching the cellular
scale. Nine normal glandular tissue samples were collected
from the peripheral zone of six formalin fixed radical prostatectomy specimens. Fibromuscular stromal tissue exhibited
microscopic diffusion anisotropy (mean fractional anisotropy
range 0.47–0.66) significantly higher (P < 0.01, Student’s t-test)
than in epithelium-containing voxels (mean fractional anisotropy range 0.31–0.54) in six of the seven normal tissue samples in which both compartments could be measured. Fiber
tracking demonstrated principle stromal fiber directions consistent with myocyte orientation seen on light microscopy of
the same sample. Diffusion tensor microimaging may be valuable for investigation of variable results from attempts to
measure diffusion anisotropy in the prostate in vivo. Magn
C 2012 Wiley Periodicals, Inc.
Reson Med 000:000–000, 2012. V
Key words: diffusion; microimaging; prostate; anisotropy;
stroma; tractography
ADC has been posited to be consistent with both the loss
of high ADC lumenal and ductal spaces and increased
cell density characteristic of prostatic adenocarcinoma
(4), however, recent evidence from diffusion microimaging studies of formalin fixed prostate tissue suggest that
distinct diffusivity differences between the epithelial
cells, the stromal matrix, and the acinar lumena are
likely to contribute to changes in ADC measured at low
spatial resolution in vivo (5–7).
Measurements of diffusion anisotropy in the prostate in
vivo have produced equivocal results with widely differing
FA values for similar tissue and no consistent correlation
between pathology and FA (8–12). However, a study of formalin fixed radical prostatectomy specimens, performed at
4.7 T with spatial resolution 0.5 0.5 0.5 mm3,
obtained diffusion anisotropy data consistent with gross
tissue architecture (13). High FA was observed in regions
of primarily fibromuscular stromal tissue with the primary
diffusion axis parallel to the assumed main fiber axis.
The study reported here seeks to investigate the potential of diffusion microimaging of fixed tissue samples to
clarify the biophysical basis of diverse findings from anisotropy measurements of the prostate in vivo.
An imaging method that generates contrast based on microscopic tissue structural properties would be expected
to provide both sensitive and specific cancer detection if
the image contrast can be made to reflect the structures
that define cancer. Diffusion-weighted water imaging
(DWI) is an obvious candidate for this purpose because
the free diffusion of water in tissue is known to be constrained by intra- and extracellular structures and cell
walls. DWI can reveal both the scale and orientation of
tissue structure because contrast depends on the net displacement of water over a specific time period in a specific direction. Two parameters are commonly used to
describe the rate and relative directional freedom of
water diffusion. These are the apparent self diffusion
coefficient (ADC) or diffusivity, and the fractional anisotropy (FA) (1), respectively.
DWI studies of prostate tissue in vivo have demonstrated a decrease in the measured ADC in cancer tissue
that correlates with Gleason grade (2,3). The decrease in
METHODS
Tissue Collection
All tissue samples were collected from radical prostatectomy specimens with institutional ethics approval and
written informed consent from tissue donors. Nine samples of normal tissue were collected from the left and/or
right lateral peripheral zone of the prostates of six
patients. Whole organs, immersed 72 h in 10% neutral
buffered formalin post surgery, were sectioned for routine
histopathology. Four-mm thick transverse slices were
examined by a specialist urologic pathologist and full
thickness tissue samples obtained with a 3 mm core
punch (sample volume 28 mm3). The selection of
regions for sampling was based on visual assessment of
the likely tissue type. In this preliminary study we
focused on samples of normal glandular tissue. Cores
were placed in vials of neutral buffered formalin and
stored 1–2 weeks at room temperature before MR imaging.
1
Discipline of Medical Radiation Sciences, Faculty of Health Sciences,
University of Sydney, Lidcombe, Australia.
2
Center for Advanced Imaging, University of Queensland, Brisbane, Australia.
3
Department of Urology, Royal Prince Alfred Hospital, Sydney, Australia.
4
Department of Anatomical Pathology, Royal Prince Alfred Hospital, Sydney,
Australia.
*Correspondence to: Roger M. Bourne, Ph.D., Discipline of Medical
Radiation Sciences, Faculty of Health Sciences, University of Sydney, PO
Box 170, Lidcombe, 1825 Australia. E-mail: [email protected]
Received 9 August 2011; revised 29 December 2011; accepted 30
December 2011.
DOI 10.1002/mrm.24179
Published online in Wiley Online Library (wileyonlinelibrary.com).
C 2012 Wiley Periodicals, Inc.
V
MR Microimaging
Tissue cores were transferred from neutral buffered formalin to phosphate buffered saline containing 0.2% v/v
gadolinium contrast agent (Magnevist, Schering AG, Germany. Dimeglumine gadopentetate 0.5 mg/mL.) giving
1
2
Bourne et al.
Table 1
Fractional Anisotropy of Tissue Compartments
Sample
1
2
3
4
5
6
7
8
9
Allc
Patient
1
2
2
2
3
4
4
5
6
Epithelial FA 6 SD (No. voxels)a
0.31
0.50
0.49
0.54
b
0.14 (327)
0.19 (472)
0.20 (176)
0.19 (124)
–
0.54 6 0.22 (134)
0.45 6 0.16 (481)
–
0.41 6 0.23 (238)
0.46 6 0.08
6
6
6
6
Stromal
Ductal
0.47 6 0.16 (283)
0.58 6 0.19 (1484)
0.53 6 0.16 (500)
0.60 6 0.17 (804)
0.64 6 0.23 (769)
0.57 6 0.20 (1079)
0.58 6 0.18 (1753)
0.63 6 0.18 (538)
0.66 6 0.16 (255)
0.58 6 0.06
0.30 6 0.11 (95)
0.45 6 0.30 (132)
0.38 6 0.15 (416)
–
0.38 6 0.14 (154)
–
0.35 6 0.14 (193)
–
0.32 6 0.13 (65)
0.36 6 0.05
a
The ‘‘epithelial’’ compartment is comprised of epithelium-containing voxels also containing unknown partial volumes of stroma and
duct.
b
Bold values represents the epithelial FA significantly differing from stromal FA (P < 0.01, Student’s t-test).
c
Mean and SD of sample means.
[Gd] ¼ 0.16 mM, and stored overnight at room temperature to wash out formaldehyde. The addition of contrast
agent reduces the sample T1 to 500 ms and enabled
faster imaging by reduction of TR. Cores were glued (cyanoacrylate ‘‘Superglue’’) to a plastic strip to constrain the
position of the sample during imaging. The core and
plastic strip were inserted into a 5 mm diameter NMR
tube filled with contrast/phosphate buffered saline solution for imaging.
Imaging was performed at room temperature (22 C) on a
Bruker (Germany) AV700 magnetic resonance microimaging
system (16.4 T vertical bore magnet interfaced to an
AVANCE II spectrometer running Paravision 5, using a
5 mm solenoidal RF coil and Micro2.5 gradient set). For DWI
a 3D spin echo DTI sequence with the following parameters
was used: TR ¼ 500 ms, TE ¼ 18 ms, number of averages ¼ 1,
FOV ¼ 8 4.5 4.5 mm, acquisition matrix ¼ 200 112 112 (data resolution ¼ 40 40 40 mm3). Diffusion parameters: d ¼ 2 ms, D ¼ 12 ms, b ¼ 1000 s/mm2, with six noncollinear directions and two b ¼ 0 images. Total imaging time ¼
14 h. The intrinsic signal-to-noise ratio (SNR) calculated
from b ¼ 0 images was 24 (14).
ROI Selection
Voxel calculations were based on selection of imaging
slices that most clearly permitted unequivocal selection
of a large number of voxels composed primarily of a single compartment type (epithelium, stroma, or duct) (7).
Adjacent slices were checked to minimize partial volume
effects. A binary mask was made from the DWI slice and
then used to select pixels from the FA data of the same
slice. As the thickness of normal epithelium is variable
(from a minimum of 15 mm), and may include infolding, selected ‘‘epithelial voxels’’ may contain significant
partial volumes of both ductal space and stromal tissue.
In three of the normal tissue samples significant regions
of ductal and/or epithelium-containing voxels could not
be confidently selected.
Histopathology
The normal glandular status of samples was confirmed
by histopathologic examination of the tissue immediately
surrounding the sample core void. Following MR microimaging one of the tissue cores was embedded and sectioned for light microscopy and immunohistochemical
staining.
Analysis
Diffusion parameters were calculated using the program
Diffusion Toolkit (www.trackvis.org, Ruopeng Wang and
Van J. Wedeen. TrackVis.org, Martinos Center for Biomedical Imaging, Massachusetts General Hospital). The
resulting image data were displayed and analyzed with
MIPAV (Version 0.4.4. mipav.cit.nih.gov, Centre for Information Technology, NIH) and Matlab (Natick). Fiber
tracks were generated with Diffusion Toolkit (FACT
propagation algorithm, angle threshold 60 , spline-filtered, DWI mask image, FA cutoff 0.1) and displayed
using TrackVis (www.trackvis.org). In Diffusion Toolkit
FA is calculated from the eigenvalues (l1, l2, l3) of the
diffusion tensor as follows:
rffiffiffi
1
FA ¼
2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðl1 l2 Þ2 þ ðl2 l3 Þ2 þ ðl3 l1 Þ2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
l1 2 þ l2 2 þ l3 2
RESULTS
Measurements of voxel FA based on ROI selection (7)
are summarized in Table 1. The mean FA in stromal voxels was significantly higher than the mean FA of epithelium-containing voxels in the same sample in six of the
seven samples in which both compartments could be
confidently distinguished. As the ductal spaces were
filled with phosphate buffered saline solution in which
diffusion would be expected to be isotropic the ‘‘ductal’’
compartment FA measurements can be considered to
represent background noise. The level of noise affecting
estimates of epithelial and stromal FA would be lower
than that seen in the ducts due to their lower diffusivity
(Fig. 2a and Ref. 7) and consequent higher SNR in the
b ¼ 1000 s/mm2 DW images.
Figure 1 shows light microscopy of sections of one of
the samples of normal glandular tissue. The normal
fibromuscular stromal tissue is characterized by a
Diffusion Anisotropy in Prostate Tissue
3
FIG. 1. Light microscopy features of normal glandular prostate tissue. a: Hematoxylin and eosin stained transverse section though a
3 mm diameter normal tissue core. Corresponding MR images are shown in Fig. 2. The section shows both normal glands (NG)
and degenerate glands with atrophic epithelium (AE). Many glands contain corpora amylacea (CA) which are precipitates of secreted
protein. b: Immunohistochemical stains. Smooth muscle actin (red) and cytokeratin (CK, brown) stain of the tissue slice adjacent to that
of image (a).
distinct heterogeneity of myocyte cell orientations (Fig.
1a) with parallel orientation persisting only on a submillimeter scale. A similar heterogeneity of actin fiber
orientation is evident in the smooth muscle actin stained
adjacent section (Fig. 1b).
The diffusion-based properties of the same tissue core,
imaged before light microscopy, are shown in Fig. 2. As
FIG. 2. Diffusion anisotropy and fiber
tractography in normal glandular tissue. a: Diffusion-weighted image of
transverse slice through 3 mm diameter tissue core. The image slice is
close to, but not exactly coplanar with,
the histology slices shown in Fig. 1. b:
Fractional anisotropy. c: Directionally
color-encoded FA (Pixel intensity proportional to FA). Color represents
direction of primary eigenvector. d:
Fiber tracks that traverse the same tissue slice as images (a)–(c). The slice is
tilted to enhance visibility of fibers
crossing the slice. Color represents
direction of fiber. Note the correspondence with principle eigenvector direction in image (c).
reported previously (6,7) epithelium-containing voxels
show low diffusivity relative to stromal tissue and ducts
(Fig. 2a). This low diffusivity correlates with high cytokeratin density (Fig. 1b). Note that the correlation
between light microscopy and MR images is imperfect as
it is not possible to section the embedded tissue in
planes exactly parallel to the MR imaging plane.
4
Bourne et al.
tracks are mainly confined to the regions of stromal tissue between glands and show minimal extension into
the epithelium layer and ductal space.
Figure 4 illustrates the heterogeneous fiber track orientation found in three further cores of normal glandular
tissue. In all samples the primary direction and spatial
location of fiber tracks were robust to changes of propagation algorithm (FACT; 2nd order Runge-Kutta; interpolated streamline; tensorline), FA cutoff, and threshold
angle. These variables mainly affected the length of generated tracks.
DISCUSSION
FIG. 3. Diffusion and fiber tracks in highly glandular tissue. a: Diffusion-weighted image illustrating multiple glands lined with lowdiffusivity epithelium. b: Fiber tracks that traverse the same tissue
slice as image (a). c: Overlay of fiber tracks onto diffusionweighted image. Note that the generated tracks are primarily confined to stromal tissue. [Color figure can be viewed in the online
issue, which is available at wileyonlinelibrary.com.]
The FA image (Fig. 2b) suggests that the highest voxel
anisotropies occur in stromal tissue and the principle
eigenvector direction image (Fig. 2c) suggests that there
are distinct local regions in which adjacent voxels have
similar primary diffusion directions.
Fiber tractography (Fig. 2d) produces fibers (streamlines) with orientation consistent with the long axis of
stromal cells seen in the H&E stained section (Fig. 1a)
and actin filaments of the adjacent section (Fig. 1b).
DWI and fiber tracks in a second normal tissue sample
are shown in Fig. 3. This sample is more densely glandular than the sample of Figs. 1 and 2. The generated fiber
This preliminary investigation provides new information
about diffusion anisotropy in normal glandular prostate
tissue.
There was considerable heterogeneity of principle diffusion direction throughout each tissue sample. The most
robust demonstration of locally homogeneous diffusion
anisotropy in stromal tissue comes from the consistency
of fiber tracking results regardless of track propagation
algorithm, FA cutoff, and angle threshold. This heterogeneous microscopic anisotropy would most likely give rise
to low ensemble anisotropy (13,15) when FA is measured
in vivo at spatial resolutions where the voxel volume is
typically similar to the tissue sample size used in this
study (28 mm3). It is possible that microscopic heterogeneity of stromal fiber direction has contributed to diverse
findings from attempts to measure FA in the prostate in
vivo (8–12). Heterogeneity of microscopic anisotropy
would also explain the generally lower FA reported at low
spatial resolution in vivo compared with FA measured in
the same organ at high spatial resolution ex vivo (13).
Noise due to the attenuation of the diffusion-weighted
signal is a well-known problem in the estimation of diffusion anisotropy (16). We estimated the intrinsic SNR
in our b ¼ 0 images to be 24 according to the method
of Dietrich et al. (14). This exceeds the level of 20 suggested by Mukherjee et al. (16) to provide reliable DTI
parameters. The good agreement between principle
eigenvector direction, streamline (fiber track) direction,
and stromal myocyte orientation seen on light microscopy of the same tissue section, together indicate that
the SNR in stromal tissue was adequate for reliable diffusion anisotropy analysis.
Due to noise, the measured FA values are likely to be
overestimates of the true tissue FA in both stromal and
epithelial compartments (16). Despite much smaller
voxel size (0.000064 mm3) compared with the fixed tissue study of Xu et al. (13) (0.125 mm3), we did not
observe higher FA. The mean stromal FA of our samples
(0.58 6 0.06) was in the upper range of the values Xu
et al. reported for ‘‘stromal BPH’’ (benign prostatic hyperplasia). From our fiber tracking images it is clear that,
although fiber direction is heterogeneous, there a many
large bundles of fibers such that a voxel of dimension
0.5 0.5 0.5 mm3 could be comprised of approximately parallel stromal fibers and would then be
expected to have a measured FA similar to the stromal
maximum we measured. Our maximum FA value of 0.66
is similar to that reported from a study of formalin-fixed
Diffusion Anisotropy in Prostate Tissue
5
FIG. 4. Fiber tracks generated from
three further normal glandular tissue
samples. Note the heterogeneity
of fiber directions which would lead
to low ensemble anisotropy at the
spatial resolution of typical in vivo
examinations. [Color figure can be
viewed in the online issue, which is
available at wileyonlinelibrary.com.]
rat heart (FA 0.62) (17) and thus could be considered
the maximum FA that can be expected in tissue comprised of densely packed muscle fibers. The maximum
FA value (0.72) that Xu et al. reported is possibly
higher than this maximum muscle FA due to noise.
The FA measured in epithelium-containing voxels was
generally lower than in stromal voxels but varied
between samples. This variation is possibly due to variable partial volume effects as many voxels spanning the
epithelial cell layer (minimum thickness 15 mm estimated on light microscopy) could be expected to contain
as little as 40% ‘‘epithelial’’ cells (A distinction between
epithelial and basal cells at this stage is not possible).
Low FA in epithelium-containing voxels relative to stromal voxels may be the result of shorter average cell
length or the absence of long actin filaments.
Although measurement of voxel mean diffusivity was
not the primary focus of this report, the coincidence of
high cytokeratin density seen on light microscopy and
low diffusivity in the epithelial cell layer seen in DWI is
noteworthy. The possibility that a dense intracellular cytokeratin matrix gives rise to hindered or restricted diffusion in epithelial cells warrants further investigation (18).
Limitations
The observations of this study are based on imaging of formalin fixed normal tissue. Water diffusivity in vivo is generally higher than in fixed tissue and it is possible that diffusion anisotropy may be altered or exaggerated by the
fixation process. Although earlier low spatial resolution
comparisons of fresh and fixed prostate tissue suggest that
relative changes in anisotropy are minor (13), more extensive studies of both normal and cancer tissue will be
required to unequivocally establish a relationship between
diffusion anisotropy observed in fixed tissue and in vivo.
CONCLUSIONS
Diffusion tensor microimaging of a limited sample of
normal glandular prostate tissue suggests that there are
significant FA differences between stromal and epithelial
tissue. Fiber tracks generated from diffusion tensor data
from one sample were consistent with stromal myocyte
and actin fiber orientation seen on subsequent light
microscopy of the same sample. Diffusion tensor microimaging may be valuable for investigation of variable
results from attempts to measure diffusion anisotropy in
the prostate in vivo.
ACKNOWLEDGMENTS
This work was supported by an Australian National
Imaging Facility access grant. The authors thank Ms
Robyne Soper for preparation of the samples for light
microscopy.
REFERENCES
1. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimental
design and data analysis—a technical review. NMR Biomed 2002;15:
456–467.
2. Vargas HA, Akin O, Franiel T, Mazaheri Y, Zheng J, Moskowitz C,
Udo K, Eastham J, Hricak H. Diffusion-weighted endorectal MR imaging at 3 T for prostate cancer: tumor detection and assessment of
aggressiveness. Radiology 2011;259:775–784.
3. Turkbey B, Shah VP, Pang X, Bernardo M, Xu S, Kruecker J, Locklin
J, Baccala AA, Rastinehad AR, Merino MJ, Shih JH, Wood BJ, Pinto
PA, Choyke PL. Is apparent diffusion coefficient associated with clinical risk scores for prostate cancers that are visible on 3-T MR
images? Radiology 2011;258:488–495.
4. Gibbs P, Liney GP, Pickles MD, Zelhof B, Rodrigues G, Turnbull LW.
Correlation of ADC and T2 measurements with cell density in prostate cancer at 3.0 tesla. Invest Radiol 2009;44:572–576.
5. Bourne R, Kurniawan N, Cowin G, Chowdhury S, Stait-Gardner T,
Sved P, Watson G, Price W. Biexponential diffusion decay in formalin fixed prostate tissue: preliminary findings. Magn Reson Med
2011. DOI: 10.1002/mrm.23291.
6. Bourne R, Kurniawan N, Cowin G, Sved P, Watson G. 16T diffusion
microimaging of fixed prostate tissue: preliminary findings. Magn
Reson Med 2011;66:244–247.
7. Bourne R, Kurniawan N, Cowin G, Sved P, Watson G. Microscopic
diffusivity compartmentation in formalin fixed prostate tissue. Magn
Reson Med 2011. DOI: 10.1002/mrm.22778.
8. Haker SJ, Szot Barnes A, Maier SE, Tempany CM, Mulkern RV. Diffusion Tensor Imaging for Prostate Cancer Detection: Preliminary Results
from a Biopsy-Based Assessment. In: Proceedings of the 13th Annual
Meeting of ISMRM, Miami Beach, Florida, USA, 2005. p 2126.
9. Sinha S, Sinha U. In vivo diffusion tensor imaging of the human
prostate. Magn Reson Med 2004;52:530–537.
10. Reinsberg SA, Brewster JM, Payne GS, Leach MO, deSouza NM. Anisotropic Diffusion in Prostate Cancer: Fact or Artifact? In: Proceedings of
the 13th Annual Meeting of ISMRM, Miami Beach, Florida, USA, 2005.
p 269.
6
11. Gurses B, Kabakci N, Kovanlikaya A, Firat Z, Bayram A, Ulug AM,
Kovanlikaya I. Diffusion tensor imaging of the normal prostate at 3
tesla. Eur Radiol 2008;18:716–721.
12. Gibbs P, Pickles MD, Turnbull LW. Diffusion imaging of the prostate
at 3.0 tesla. Invest Radiol 2006;41:185–188.
13. Xu JQ, Humphrey PA, Kibel AS, Snyder AZ, Narra VR, Ackerman
JJH, Song SK. Magnetic resonance diffusion characteristics of histologically defined prostate cancer in humans. Magn Reson Med 2009;
61:842–850.
14. Dietrich O, Raya JG, Reeder SB, Reiser MF, Schoenberg SO. Measurement of signal-to-noise ratios in MR images: influence of multichannel coils, parallel imaging, and reconstruction filters. J Magn Reson
Imaging 2007;26:375–385.
Bourne et al.
15. Shemesh N, Cohen Y. Microscopic and compartment shape anisotropies in gray and white matter revealed by angular bipolar doublePFG MR. Magn Reson Med 2011;65:1216–1227.
16. Mukherjee P, Chung SW, Berman JI, Hess CP, Henry RG. Diffusion
tensor MR imaging and fiber tractography: technical considerations.
Am J Neuroradiol 2008;29:843–852.
17. Wu Y, Wu EX. MR study of postnatal development of myocardial
structure and left ventricular function. J Magn Reson Imaging 2009;
30:47–53.
18. Assaf Y, Freidlin RZ, Rohde GK, Basser PJ. New modeling and
experimental framework to characterize hindered and restricted
water diffusion in brain white matter. Magn Reson Med 2004;52:
965–978.