Clinical Oncology

Whole-Body Diffusion-Weighted MRI of

the Bone Marrow in Health and Disease
Anwar R. Padhani1; Hassan Douis2; Peter Gall3
1
Paul Strickland Scanner Centre, Mount Vernon Cancer Centre, Northwood, Middlesex, UK
2
Department of Radiology, Royal Orthopaedic Hospital, Birmingham, UK
3
Siemens Healthcare, Erlangen, Germany

Background tracers include 18F-flurodeoxyglucose cancer) and prostate specific membrane

A comprehensive imaging assessment of (FDG) for glucose metabolism, 11C/18F- antigen (PSMA) [2]. Dynamic contrast-
the skeletal bone marrow requires a thymidine (FLT) for DNA synthesis, enhanced MRI (DCE-MRI) can evaluate
multimodal approach because no one 11
C/18F-acetate for fatty acid synthesis, the vascularisation of bone marrow and
technique is able to evaluate all aspects 11
C-methionine for protein synthesis its cellular content can be assessed
of bone health. Thus, 99mTc-MDP skeletal and 11C/18F-choline for cell membrane using Dixon and diffusion MRI tech-
scintigraphy and 18F-Na PET evaluate synthesis and degradation [1]. Tumor niques. These techniques have variable
osteoblastic function, ultrashort TE and specific tracers include 18F-FES (fluoro- clinical availability and whole-body
CT scans evaluate the structural proper- estradiol for breast cancer), 18F-FDHT imaging capability.
ties of bone and the osteolytic activity of (fluorodihydrotestosterone for prostate In recent years, whole-body MRI has
osteoclasts. Non-specific PET tracers can cancer) and engineered antibody frag- emerged as an excellent imaging modal-
evaluate bone marrow/tumor membrane ments that target tumor cell-surface ity for the evaluation of normal bone, of
transporters and metabolism. Relevant targets such as HER-2/neu (for breast bone marrow infiltration and the detec-

1
b50 b900 ADC B900 coronal MPR Inverted 3D MIP

1 WB-DWI workflow. 27-year-old-woman with sarcomatoid left breast cancer (arrow). The bone marrow pattern is normal for this age. Axial
DWI from the skull base to the mid-thigh is performed using 2 b-values (50 and 900 s/mm2) with a slice thickness of 5 mm in 4 stations. The
b900 images are reconstructed into the coronal plane (5 mm) and displayed as thick 3D MIPs (inverted grey scale). ADC images are computed
inline with mono-exponential fitting of b50 and b900 signal intensities.

60 MAGNETOM Flash · 2/2012 · www.siemens.com/magnetom-world

 Oncology Clinical

tion of skeletal metastases due to its 2

exquisite soft tissue contrast, high spa-
tial resolution and lack of ionizing radia-
tion [3]. Whole-body MRI is particularly
attractive for children* and young adults
because it lacks ionizing radiation.
Despite its high sensitivity and high
specificity for the detection of skeletal
metastases, whole-body MRI has not
been widely adopted into daily clinical
practice. This is predominantly due to
the fact, that whole-body MRI performed
with conventional sequences has to be
tailored for each tumor type, has long
examination times, often requires the
administration of intravenous contrast
and importantly, it is time consuming to
analyse and report [4]. However, there is
a strong clinical need for whole-body
tumor assessments particularly for can-
cer patients.
Whole-body diffusion-weighted MRI
(WB-DWI) is a new powerful adjunct to
anatomical whole-body MRI because it
provides a functional assessment of dis-
ease burden, can quantify disease
extent, does not require the administra-
tion of exogenous contrast medium and
can be performed in reasonably short
examination times. Furthermore, WB-
DWI can improve the reading and test
performance of anatomical whole-body
MR examinations because areas of
increased cellularity are depicted as
regions of high signal-intensity, there-
fore allowing an ‘at-a-glance’ assessment
of disease burden and distribution [5].
Whole-body DWI is emerging as a partic-
ularly promising imaging technique in
the detection, assessment of extent and 2 Bone marrow hypoplasia due to chemotherapy with disease progres-
therapy monitoring of skeletal metasta- sion. 49-year-old-woman with metastatic breast cancer before and after
ses [6] because it is sensitive to bone 3 cycles of carboplatin chemotherapy. Both rows left-to-right: spine
T1-weighted spin-echo, spine T2-weighted spin-echo with spectral fat
marrow cellular density, the relative pro-
saturation and b900 3D MIP (inverted scale) images. Top row before che-
portion of fat and marrow cells, water motherapy shows normal background bone marrow pattern with super-
content and bone marrow perfusion. imposed small volume bone metastases (arrow heads). Bottom row after
chemotherapy shows marked disease progression with enlarging and
Technique of whole-body DWI new bony metastases (arrows). Note that bone marrow hypoplasia has
developed in the ribs, spine and pelvis. Note reductions of signal inten-
Most modern high-field MRI systems
sity of the spleen secondary to iron deposition due to blood transfu-
possess echo-planar and parallel imag- sions. There is a right sided silicone containing breast enhancement bra
ing capabilities and allow the use of pad in place on both examinations.
high-performance gradients which
together with phased-array multichan-
nel surface coils, make it feasible for
*MR scanning has not been established as safe for imaging fetuses and infants
WB-DWI to be implemented into clinical under two years of age. The responsible physician must evaluate the benefit of
practice. Currently we find that WB-DWI the MRI examination in comparison to other imaging procedures.

MAGNETOM Flash · 2/2012 · www.siemens.com/magnetom-world 61

3

Clinical Oncology

is best performed at 1.5T using multiple

surface coils for signal reception which
allows uniform fat-suppression over
large fields-of-view. Although imaging at
3T increases the signal-to-noise ratio,
WB-DWI remains challenging because of
increased susceptibility artifacts and
poorer fat suppression.
A multiple-averaged, free-breathing
technique is most commonly used for
data acquisition. In general, the use of
two diffusion sensitizing gradients is
sufficient to enable good quality, clini-
cally useable images to be acquired. Low
3 Bone marrow hypoplasia due to aromatase inhibitor therapy. 56-year-old
b-value black-blood (50 s/mm2) com-
post-menopausal woman with estrogen receptor positive breast cancer on long
term anastrozole therapy. Long term therapy of aromatase inhibitors such as
bined with high b-value (800–1000 s/
anastrozole put women at greater risk for of osteoporosis which in turn leads to mm2) images are used in combination
increased bone marrow fat. Left-to-right: spinal T1-weighted GRE sequence for anatomic depiction, disease detec-
(from the scout views), T1-weighted spin-echo, T2-weighted spin-echo with tion and for the calculation of apparent
spectral fat saturation and b900 3D MIP (inverted scale) images. There is some
diffusion coefficient (ADC) values [5, 7].
loss of vertebral height of the D9 vertebral body. Note the visibility of the entire
length of the spinal cord on the 3D inverted MIP images and ready visibility of Our preferred method for fat suppres-
the dorsal nerve root ganglia of the brachial and lumbar plexuses. Focal high- sion uses inversion recovery because it
signal lesions over the left chest are due to an acneiform rash on the skin. allows uniform fat suppression over
large fields-of-view [8].
Our institutional protocol combines
4A 4B
whole-body MRI and whole-body DWI
done on a 1.5T MAGNETOM Avanto
scanner (Siemens Healthcare, Erlangen,
Germany) equipped with a continuous
moving table option and total imaging
matrix (Tim) body surface coils. First,
conventional sagittal T1-weighted and
T2-weighted fat-suppressed images of
the spine are obtained. Subsequently,
axial T1-weighted and STIR images from
the skull vault to the mid-thigh are
acquired using continuous table move-
ment employing multiple breath-holds
for image acquisitions of the chest,
abdomen, pelvis and upper thighs. Axial
DWI from the skull vault to the mid-thigh
is then performed using b-values of 50
s/mm2 and a b-value of 900 s/mm2 with
a slice thickness of 5 mm. The axial DWI
acquisition is usually achieved in 4 con-
tiguous stations using a free-breathing
technique, with each station taking
approximately 6 minutes to acquire. The
high b-value images are then recon-
structed in orthogonal planes as thin
4 Bone marrow hyperplasia induced by chemotherapy with G-CSF therapy.
multiplanar reconstructions (5 mm) and
50-year-old woman with metastatic breast cancer before and after 3 cycles of
erubulin chemotherapy with growth-colony stimulating factor (G-CSF) given to
as thick 3D maximum intensity projec-
prevent neutropenia. b900 3D MIP (inverted scale) images. Left image show tions (MIPs) which are usually displayed
multiple bone metastases (arrows). Right image after 3 cycles of chemotherapy using an inverted grey scale. ADC maps
shows increases in signal intensity of the bone marrow leading to the are computed inline with system soft-
decreased visibility of the bone metastases. The splenic size has also increased.
The increased signal intensity of the background bone marrow should not be
misinterpreted as malignant progression.
 Oncology Clinical

ware using mono-exponential fitting in 5A 5B 5C 5D

which each voxel reflects the tissue dif-
fusivity (units: µm2/s) (Fig. 1).

Bone marrow imaging

on WB-DWI
In order to be able to utilize WB-DWI in
the detection, characterization and
treatment assessment of skeletal metas-
tases, it is vital to be familiar with the
signal distribution of normal bone mar-
row on high b-value images. Normal
adult bone marrow distribution becomes
established by the age of 25 years with
mixed red bone marrow found in the
axial skeleton, the humeral and femoral
proximal metaphyses whilst yellow mar-
row is found in the appendicular skele-
ton. Thereafter, conversion of red mar- 5 Poor visibility of treated metastases and osteoblastic metastases. 69-year-old
row regions to yellow marrow occurs at man with metastatic prostate cancer on long term, third line hormonal therapy with
a slow rate, the speed of conversion abiraterone being evaluated for rising serum prostate specific antigen (PSA) levels.
being dependent on patient gender and He has had an excellent response to 2 years of treatment with residual abnormalities
in his bone marrow visible on T1-weighted (5A) and T2-weighted (5B with fat sup-
underlying medical conditions [9]. It is
pression) spinal images. No hyperintensity is seen on the b900 3D MIP (inverted
thought that the reduced water content scale) image (5C) indicating the absence of osteolytic disease. Bone scan (5D) shows
[9], the larger-sized fat cells, the hydro- a focal area of osteoblastic uptake in the intertrochanteric region of the left femur
phobic nature of fat and poorer perfu- (arrow) which is not visible as a discrete region on the b900 3D inverted MIP image.
sion all contribute to lower signal inten-
sities and ADC values of the yellow bone
marrow. On the other hand, with
increasing cellularity and water content WB-DWI, bone marrow hypocellularity that WB-DWI should not be performed
and greater perfusion, mixed yellow-red demonstrates decreased signal intensity nor interpreted in isolation but rather
bone marrow returns higher signal on high b-value images with the has to be considered as a valued adjunct
intensities and paradoxically higher ADC increased visibility of the spinal cord to anatomical whole-body MRI and
values [5, 10–12]. The changing distri- (Fig. 3). Marrow hypercellularity pres- therefore needs to be interpreted in con-
bution of the normal marrow is exqui- ents as diffuse increases in signal inten- junction with conventional WB-MRI stud-
sitely demonstrated on WB-DWI (Fig. 1). sity on high b-values [5, 6]. Changes ies [13]. This assertion has been high-
Similarly, both bone marrow hypo- and in background bone marrow cellularity lighted in a recent meta-analysis which
hypercellularity are also well depicted on can affect the visibility of bone marrow demonstrated that the high sensitivity of
WB-DWI. Common causes for bone mar- metastases. Bony metastases can WB-DWI to detect metastases was at the
row hypocellularity include chemother- become less conspicuous against expense of specificity [3]. Causes for
apy (Fig. 2), radiotherapy, myeloprolifer- increasing background signal intensities false-positive findings are bone marrow
ative disorders (e.g. myelofibrosis, when growth colony stimulating factor edema caused by fractures, osteoarthri-
myelodysplasia), non-malignant marrow (G-CSF) is used to prevent chemotherapy tis, infection, bone infarcts, vertebral
disorders (e.g. aplastic anemia), old induced neutropenia (Fig. 4). On the hemangiomas, isolated bone marrow
age and osteoporosis (including drug other hand, the detection of bone islands and bone marrow hyperplasia
induced), chronic disease (e.g. renal fail- metastases is improved in bone marrow due to G-CSF (Fig. 4). However, many of
ure, chronic liver disease, rheumatoid that is relatively hypocellular, for exam- these false-positive findings can be over-
arthritis) and prolonged immobility. In ple in the older patient or after chemo- come by correlating high b-value DW
contrast, relative bone marrow hyper- therapy (Fig. 2). images with ADC maps and anatomical
cellularity is observed in children and sequences [5].
adolescents, chronic anaemia, in smok- Skeletal metastases detection In contrast, causes for false-negative
ers, chronic cardiac failure, in pregnancy On WB-DWI, skeletal metastases appear findings are low levels of bone marrow
and in patients treated with hematopoi- as focal or diffuse areas of high signal infiltration such as in smoldering multi-
etic growth factors such as granulocyte- intensity on high b-values (Figs. 2, 4, 5). ple myeloma or when background bone
colony stimulating factors (G-CSF). On It is, however, important to emphasize marrow hyperplasia obscures the pres-

MAGNETOM Flash · 2/2012 · www.siemens.com/magnetom-world 63

Clinical Oncology

ence of metastases. Similarly, the detec- water and cellular content of sclerotic diffusivity. On WB-DWI, this leads to
tion of skeletal metastases on WB-DWI and treated metastases [10, 14]. Diffu- marked increases of ADC values. The
may be impaired in areas of macroscopic sion-weighted MRI is better at detecting extent of ADC increases after successful
movement such as the anterior ribs and skeletal lesions from smaller cancer cell therapy is thought to be related to
sternum. Visibility of skull vault infiltra- size infiltrations such as those due to mechanism of tumor cell death. It has
tions can be impaired because of the breast cancer, myeloma, lymphoma and been proposed that therapies which result
adjacent high signal of the brain. The small cell tumors such as neuroendocrine in necrotic cell death may result in higher
visibility of skull base disease is impaired tumors compared to bony metastases ADC values than treatment regimes
because of susceptibility effects. Another from clear cell cancers of the kidneys. which result in non-necrotic tumor cell
important cause for false-negative find- death. This is thought to be due to the
ings is successfully treated malignant Therapy assessment of skeletal fact that necrotic cell death is associated
disease and sclerotic metastases (Fig. 5). metastases on WB-DWI with an inflammatory response which
In general, lytic lesions are better Successful treatment of skeletal metas- results in increased tissue water. Signal
detected than sclerotic or treated lesions tases is characterized by tumor cell intensity changes on high b-values
on WB-DWI. This is due to the lower death which results in increased water images after successful therapy are more

6A 6B

6 Therapy response in metastatic breast cancer. Serial changes in a 64-year-old woman with metastatic breast cancer responding to treatment
with FEC (Fluorouracil (5FU), epirubicin and cyclophosphamide) chemotherapy with bisphosphonates. (6A) b900 3D MIP (inverted scale) images
before (left column) and after 3 cycles of treatment (right column). A diffuse pattern of metastatic bone disease is seen pre-treatment that
decreases in extent and signal intensity indicating disease response. There are a few focal areas of persistent signal hyperintensity post-therapy
indicating likely active disease. (6B) Histogram analysis of a pelvic volume of interest defined on b900 images of the pre-treatment examination
and applied to the second examination after robust image registration. Threshold histograms of muscle normalized b900 signal intensity (top
row) and ADC values (middle row), and pixel scatter plots (bottom row) of normalized b900 signal intensity (x-axis) and ADC values (y-axis).
The control lines on the histograms and scatter plots are placed on 3 and 9 for normalized b900 signal intensity and 650 and 1500 µm2/s for
ADC values. The histograms show reductions in signal intensity (more blue pixels and less overall spread) on top row accompanied by increases in
ADC values (more green and red pixels). Note how the pixel scatter plot moves to the top left. Analyses were done using OncoTreat* software
(Works-in-progress, Siemens Healthcare, Erlangen, Germany).

*This information about this product is preliminary. The product is under development and not commercially available in the U.S., and its future availability cannot be ensured.

64 MAGNETOM Flash · 2/2012 · www.siemens.com/magnetom-world

 Oncology Clinical

varied, but in most cases decreases in sig- being used for the evaluation of patients References
nal intensity are seen (Fig. 6). with melanoma, lymphoma and renal 1. Jadvar H. Prostate cancer: PET with
Occasionally, successful treatment cancer. We have also found it valuable 18F-FDG, 18F- or 11C-acetate, and 18F- or
11C-choline. J Nucl Med 2011;52:81-89.
response may demonstrate persistent for the examination of children* and
2. Mankoff DA, Link JM, Linden HM, Sundara-
high signal intensity on high b-value young adults, pregnant* women with rajan L, Krohn KA. Tumor receptor imaging.
images but with an associated increase cancer and in patients with impaired J Nucl Med 2008;49 Suppl 2:149S-163S.
in ADC-values. This appearance is renal function or allergies when there is 3. Wu LM, Gu HY, Zheng J, et al. Diagnostic
termed ‘T2 shine-through’ and is due to a contraindication to the intravenous value of whole-body magnetic resonance
imaging for bone metastases: a systematic
tumor necrosis. It is thus prudent to injection of contrast medium.
review and meta-analysis. J Magn Reson
always correlate high b-value images Imaging 2011;34:128-135.
with ADC values when interpreting WB- Conclusions 4. Schmidt GP, Reiser MF, Baur-Melnyk A.
DWI in order to avoid misinterpreting WB-DWI has emerged as a powerful Whole-body MRI for the staging and
persistent high signal intensity on high imaging tool in the detection, character- follow-up of patients with metastasis.
Eur J Radiol 2009;70:393-400.
b-value images as non-responding dis- ization and for monitoring of treatment
5. Padhani AR, Koh DM, Collins DJ. Whole-
ease. Another pattern is a decrease in response of tumors. It particularly excels body diffusion-weighted MR imaging in
signal intensity on high b-values associ- in the detection and disease extent cancer: current status and research direc-
ated with unchanged ADC values or assessment of bone marrow lesions as tions. Radiology 2011;261:700-718.
6. Padhani AR, Gogbashian A. Bony metasta-
slight decreases in ADC values. Our well as in the evaluation of therapy
ses: assessing response to therapy with
experience from comparison with CT response of skeletal lesions. It is how- whole-body diffusion MRI. Cancer Imaging
examinations, Dixon images and ana- ever important to emphasize the fact 2011;11 Spec No A:S129-145.
tomical MRI studies suggests that this that WB-DWI has to be interpreted in 7. Padhani AR. Diffusion magnetic resonance
pattern may be due to sclerosis or fibro- conjunction with conventional whole- imaging in cancer patient management.
Semin Radiat Oncol 2011;21:119-140.
sis. We have classified this pattern as body MRI sequences in order to avoid
8. Koh DM, Blackledge M, Padhani AR, et al.
indeterminate and therefore take into pitfalls in diagnoses. There is an urgent Whole-Body Diffusion-Weighted MRI: Tips,
account other radiological and clinical need to better understand the biologic Tricks, and Pitfalls. AJR Am J Roentgenol
findings for response assessment. basis of treatment-related changes. Fur- 2012;199:252-262.
By contrast, patients who fail to respond thermore response criteria need to be 9. Hwang S, Panicek DM. Magnetic resonance
imaging of bone marrow in oncology,
to therapy usually demonstrate persis- established and tested in prospective
Part 1. Skeletal Radiol 2007;36:913-920.
tent or increasing bone marrow hyperin- clinical studies. Nevertheless, WB-DWI 10. Messiou C, Collins DJ, Morgan VA, Desouza
tensity on high b-value images. In the demonstrates great potential in the eval- NM. Optimising diffusion weighted MRI
context of disease progression, a uation of bone marrow lesions because for imaging metastatic and myeloma bone
disease and assessing reproducibility.
decrease or no change in ADC values it is quick to perform, can be incorpo-
Eur Radiol 2011;21:1713-1718.
can be observed because of increasing rated into standard clinical protocols and 11. Hillengass J, Bauerle T, Bartl R, et al. Diffu-
in tumor cell density within the confines can potentially answer clinical questions sion-weighted imaging for non-invasive
of a fixed bone marrow space. Small regarding tumor response more reliably and quantitative monitoring of bone mar-
increases in ADC values can also be than conventional imaging modalities. row infiltration in patients with monoclonal
plasma cell disease: a comparative study
observed in progressive disease (but of a
with histology. Br J Haematol
much smaller magnitude compared to *MR scanning has not been established as safe 2011;153:721-728.
responding lesions). For this reason we for imaging fetuses and infants under two years 12. Nonomura Y, Yasumoto M, Yoshimura R, et
use an upper ADC cut-off value derived of age. The responsible physician must evaluate al. Relationship between bone marrow cel-
the benefit of the MRI examination in compari- lularity and apparent diffusion coefficient.
from untreated patients in order to dif- son to other imaging procedures.
J Magn Reson Imaging 2001;13:757-760.
ferentiate responders from non-respond-
13. Lecouvet FE, El Mouedden J, Collette L,
ers. This cut-off value is likely to be et al. Can whole-body magnetic resonance
dependent on the choice of b-values imaging with diffusion-weighted imaging
utilized and the tumor type [5, 6]. replace Tc 99m bone scanning and com-
Contact puted tomography for single-step detection
Clinical indications Prof. Anwar R. Padhani, MBBS FRCP FRCR of metastases in patients with high-risk
Paul Strickland Scanner Centre prostate cancer? Eur Urol 2012;62:68-75.
At our institution, we have performed 14. Eiber M, Holzapfel K, Ganter C, et al. Whole-
Mount Vernon Cancer Centre
more than 1,200 WB-MRI examinations Rickmansworth Road body MRI including diffusion-weighted
incorporating diffusion MRI in the last Northwood imaging (DWI) for patients with recurring
prostate cancer: Technical feasibility and
four years using it for bone marrow Middlesex HA6 2RN
United Kingdom assessment of lesion conspicuity in DWI. J
metastasis detection and therapy moni- Magn Reson Imaging 2011;33:1160-1170.
Phone: +44-(0) 1923-844751
toring for a variety of cancer types – Fax: +44-(0) 1923-844600
principally in breast and prostate cancer [email protected]
and multiple myeloma. It is increasingly

MAGNETOM Flash · 2/2012 · www.siemens.com/magnetom-world 65