Clinical Radiology 70 (2015) 1116e1121

Contents lists available at ScienceDirect

Clinical Radiology
journal homepage: www.clinicalradiologyonline.net

Clinical relevance of the apparent diffusion

coefficient value of metastatic bone tumours on
diffusion-weighted MRI images: differences
according to the types of primary tumour, the
affected bones, and clinical factors
M.J. Cha, Y.C. Yoon*
Department of Radiology and Center for Imaging Science, Samsung Medical Center, Sungkyunkwan University School
of Medicine, Seoul, South Korea

article in formation AIM: To evaluate whether the apparent diffusion coefficient (ADC) of metastatic bone tu-
mours on diffusion-weighted magnetic resonance imaging (MRI) images differs according to
Article history: the type of primary cancer, the affected bone, and clinical factors.
Received 8 November 2014 MATERIALS AND METHODS: For this retrospective study, two radiologists reviewed MRI
Received in revised form images, including ADC maps, of 67 patients (M:F¼38:29; median age, 48 years) who were
13 February 2015 diagnosed with bone metastasis by means of histological or clinical confirmation. The primary
Accepted 29 May 2015 tumours included 29 lung adenocarcinomas, 15 invasive ductal adenocarcinomas of the breast,
13 hepatocellular carcinomas, six prostatic carcinomas, and four renal cell carcinomas. ADC
values of the metastatic tumour were compared according to the type of primary malignancy,
the affected bone, and the age and sex of the patient using KruskaleWallis and ManneWhitney
U-tests with Bonferroni correction. In addition, pre-contrast CT images were available in 38 of
67 patients; a subanalysis of the CT radiodensity and ADC values were performed with
Spearman correlation.
RESULTS: The mean, standard deviation, and minimum and maximum values of the ADC of
metastatic bone tumours did not differ significantly according to type of primary malignancy,
the affected bone, or clinical variables (p>0.1). The ADC value was not significantly correlated
with CT radiodensity (p¼0.24). Intra- and interobserver agreements for the mean ADC values
were excellent (intra-observer: p¼0.98; interobserver: p¼0.98).
CONCLUSIONS: Assessment of the ADC value of metastatic bone tumours is not reliable for
differentiation of the type of primary cancer.
Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

* Guarantor and correspondent: Y.C. Yoon, Department of Radiology and
Center for Imaging Science, Samsung Medical Center, Sungkyunkwan Uni-
versity School of Medicine, 81 Irwon-ro, Gangnam-gu, Seoul 135-710, South
Korea. Tel.: þ82 2 3410 6454; fax: þ82 2 3410 0084.
E-mail address: youngcheol.yoon@gmail.com (Y.C. Yoon).
Introduction
Bone metastasis is a frequent complication of cancer,
occurring in up to 70% of patients with advanced breast or
prostate cancer and in approximately 15e30% of patients
with carcinoma of the lung, gastrointestinal (GI) tract, or

http://dx.doi.org/10.1016/j.crad.2015.05.015
0009-9260/Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
 M.J. Cha, Y.C. Yoon / Clinical Radiology 70 (2015) 1116e1121 1117

kidney.1,2 It is not uncommon to encounter bone metastasis
as an initial presentation of cancer manifestation. When a
mass of the musculoskeletal area is observed in clinical
practice, imaging, especially magnetic resonance imaging
(MRI), plays an important role in the diagnosis and staging
of the lesion. Morphological parameters such as size,
margin, involvement of adjacent structures, signal homo-
geneity, and degree of enhancement on contrast-enhanced
MRI images can be used to distinguish between benign and
malignant tumours.3e5 Nevertheless, imaging usually offers
limited information about the type of cancer and histo-
pathological grade and is not a definitive diagnostic test.
Diffusion-weighted (DW) imaging is a novel quantitative
method for observation and measurement of microscopic
diffusion of water molecules in biological tissues.6 Because
diffusion properties differ between normal tissue and ma-
lignant masses, DW imaging can be used to detect malig-
nant conditions, such as primary cancer or distant
metastasis.7e9 Several studies have assessed whether the
use of the apparent diffusion coefficient (ADC) provides a
quantitative method for differentiating histological sub-
types and histopathological grade in various organs, such as
the liver, pancreas, kidney, cerebellum, and soft tissue10e14;
however, information is still limited about the clinical
application of DW imaging and ADC for the characterisation
of bone tumours, especially bone metastases.
The present authors hypothesised that the ADC will
reflect the microscopic characteristics of metastatic bone
tumours by the amount of interstitial space or cellularity,
which may differ according to the type of primary cancer,
the targeted bone, and clinical factors. The purpose of the
present study was to investigate the clinical relevance and
effectiveness of the novel MRI sequences in the character-
isation of bone metastasis. DW imaging results and the ADC
values of metastatic bone tumours were retrospectively
examined with the hypothesis that ADC values allow for
differentiation of primary cancer types.
with double primary cancer were excluded due to an un-
determined origin of the bone metastasis. Eight other pa-
tients were excluded due to image-related issues, such as a
small scan range and artefacts.
Finally, a total of 67 patients (M:F¼38:29; median age, 48
years) were included in this study. The primary tumours
were surgically verified in all patients. There were 29 lung
adenocarcinomas, 15 invasive ductal carcinomas of the
breast, 13 hepatocellular carcinomas, six prostatic carci-
nomas, and four renal cell carcinomas. The affected bones
were classified into two categories: long bones (n¼24) and
irregular bones (n¼43; Table 1). Long bones included the
humerus, femur, and tibia, and vertebrae; the sacrum,
scapula, and pelvic bones were categorised as irregular
bones.
MRI technique
All patients underwent MRI prior to initiating chemo-
therapy or radiation therapy. MRI was performed using a 3 T
MRI system (Intera Achieva 3 T, Philips Medical System,
Best, The Netherlands) and the coil that was used depended
on the scanned site (e.g., dual, cardiac, torso, or knee coil).
Conventional MRI sequences including T1-weighted (T1W)
and T2-weighted (T2W) axial and sagittal images, T1W
coronal images, and fat-saturated T2W sagittal images were
obtained. Contrast-enhanced axial and coronal T1-
weighted images were acquired after intravenous injec-
tion of contrast material (gadoteratemeglumine; Dotarem,
Guerbet, Aulnaysous Bois, France, 0.1 mmol/kg body weight
Table 1
Apparent Diffusion coefficient (ADC) value of bone metastasis in 67
consecutive patients.
No. (%) ADC values (106 mm2/s)
of Patients
Mean SD Min Max

Materials and methods Primary malignancy

Lung 29 (43.3) 785.74 89.99 555.63 996.97
adenocarcinoma
The institutional review board approved this study, and Invasive ductal 15 (22.4) 779.57 84.81 574.97 983.03
informed consent was waived for the use of patients’ adenocarcinoma
medical data (IRB-2014-04-099). of breast
Hepatocellular 13 (19.4) 711.38 101.42 447.43 984.66
carcinoma
Patients
Prostatic 6 (8.9) 753.72 111.78 492.6 1058.53
carcinoma
From an oncology database at Samsung Medical Center, Renal cell 4 (6) 988.3 155.98 563.23 1446.05
78 cancer patients were identified who underwent carcinoma
musculoskeletal MRI, including DW imaging between p-Value 0.29 0.18 1.0 0.26
Affected bone
February 2011 and November 2012 and were diagnosed Irregular bone 43 (64.2) 762.71 85.30 552.70 966.82
with bone metastasis. The diagnosis of bone metastasis was Long bone 24 (35.8) 808.61 117.79 499.87 1125.84
based on the results of either bone biopsy (n¼12) or clin- p-Value 0.36 0.001* 0.30 0.021*
icoradiological settings (n¼66). The clinicoradiological Sex
Male 38 (56.7) 792.06 106.12 517.93 1069.11
diagnosis was established with three prerequisite condi-
Female 29 (43.3) 762.25 84.9 554.54 964.39
tions: typical imaging features, progression in size and p-Value 1.0 0.8 1.0 0.31
number during the follow-up period, and therapeutic Age
response after systemic anticancer treatment. Patients had p-Value 1.0 1.0 1.0 1.0
to be older than 35 years with no history of anti-cancer Note. *: P value <.05.
treatment before the MRI was obtained. Three patients SD, standard deviation; Min, minimum; Max, maximum.
1118 M.J. Cha, Y.C. Yoon / Clinical Radiology 70 (2015) 1116e1121

by power injector) with spectral pre-saturation and inver- undergone PET/CT, the maximum standardised uptake
sion recovery for fat suppression. The section thickness value (SUVmax) of the bone metastases was also obtained.
ranged from 3 to 6 mm. Matrix size and the field of view
(FOV) varied in accordance with the anatomical location Statistical analysis
and size of the tumour.
DW images were acquired with a single-shot spin-echo A KruskaleWallis test with Bonferroni correction was
echo-planar imaging sequence in three orthogonal used to compare ADC measurements between the five
diffusion-encoding directions, with three b-values (0, 400, primary cancer groups. To investigate the association be-
and 1400 s/mm2). SENSE parallel imaging (acceleration tween the ADC value and clinical or demographic variables,
factor¼2) and spectral presaturation with inversion recov- such as the affected bone and the sex and age of the patient,
ery fat saturation were implemented to reduce image ar- one-way analysis of variance (ANOVA), ManneWhitney U-
tefacts. The DW images consisted of 20 transverse sections tests, and Pearson correlation analysis were used, all of
with a section thickness of 4 or 5 mm. The ADC maps were which were corrected with the Bonferroni correction.
automatically generated from the DW images with com- Paired t-tests were used to assess the variation of ADC value
mercial diffusion-analysis software (Extended MRI work- within a patient who had multiple bone metastases. A
space, version 2.6.3.1. Philips Healthcare). Three b-values (0, Spearman correlation test was used to assess the relation-
400, and 1400 s/mm2) were used; the b value of 1400 s/ ship between the ADC value and CT density and FDG uptake.
mm2 was chosen because a previous report found that this Intra- and interobserver agreements were assessed with
value was optimal for imaging bone marrow.15 intraclass correlation coefficients (ICCs) calculated with a
two-way random-effects model. The 95% confidence inter-
Image analysis val (CI) was calculated with the delta method after Fisher z
transformation. Statistical analysis was performed with
MRI images were independently reviewed by two radiol- SPSS version 15 for Windows (SPSS, Chicago, IL, USA).
ogists (Y.C.Y., with 11 years of experience in musculoskeletal p0.05 were considered statistically significant.
MRI; and M.J.C, a 4th-year resident in radiology) using the
picture archiving and communications system (PACS; Results
Centricity, GE Healthcare, Chicago, IL, US). The enhanced solid
portion of the bone metastasis was identified on post- Association between the type of primary cancer and the
contrast T1W images. Regions of interest (ROIs) were
ADC value
manually drawn freehand over the solid portion of the
tumour by two radiologists at the workstations. If a patient
The mean ADC value for bone metastasis did not differ
had multiple bone metastases, the largest one was selected. If
according to the type of primary cancer (p¼0.29; Figs 1
other pathological conditions, such as fracture or infection,
and 2). The mean, standard deviation, and minimum and
were suspected, the tumour was excluded in the analysis.
maximum ADC values did not differ significantly according
ROIs were drawn on the axial section with the greatest area of
to the type of primary cancer (Table 1). In terms of the ADC
viable tumour. Care was taken to exclude regions of necrotic
value within a patient who had more than one bone me-
or cystic change, haemorrhage, and oedema after thorough
tastases simultaneously, the variation of mean ADC value
review of corresponding conventional and contrast-
between two metastatic bone tumours was not significant
enhanced MRI images. The mean, standard deviation, and
maximum and minimum ADC values were obtained for each
ROI. In patients who have more than one measurable meta-
static lesion, the ROI was additionally drawn on the second
largest one to evaluate the variation of ADC value within a
patient. To assess intra-rater reliability, one of the two radi-
ologists performed a second evaluation 2 weeks later while
blinded to the results of the first evaluation.
In addition, 38 of 67 patients underwent computed to-
mography (CT) or combined positron-emission tomography
(PET)/CT of the bone metastasis within a 2-month period
from the day of MRI examination. All CT or PET/CT images
were obtained before systemic anti-cancer treatment. To
evaluate the correlation between the ADC value and the CT
radiodensity of each metastatic lesion, an ROI was manually
drawn on the pre-contrast CT images at the same lesion and
the same level. Every effort was made to place each ROI at
roughly the same location between the ADC maps and the
CT images. Measurements including the mean, standard
deviation, and minimum and maximum CT radiodensity Figure 1 Box plot showing the mean ADC for metastatic bone tu-
(HU) of the ROIs were obtained. In 17 patients who had mours in relation to the type of primary cancer.
 M.J. Cha, Y.C. Yoon / Clinical Radiology 70 (2015) 1116e1121 1119

Figure 2 ADC maps of bone metastasis from five different primary cancers. (a) Left humeral metastasis from lung adenocarcinoma; mean ADC of
833.6 mm2/s. (b) Right ischium with metastatic breast cancer; mean ADC of 911.9 mm2/s. (c) Right ischial metastasis from HCC; mean ADC of
587.9 mm2/s. (d) Right scapula with metastatic prostatic cancer; mean ADC of 613.6 mm2/s. (e) Sternal metastasis from RCC; mean ADC of
1070.3 mm2/s.

(p¼0.135). The standard deviation, maximum and mini-
mum value did not show significant variation either
(p¼0.338, p¼0.505, and p¼0.288, respectively). Intra- and
interobserver agreements between the two readers for the
measurement of ADC values were excellent: the ICC was
0.981 (95% CI: 0.969, 0.988) for intra-observer agreement
and 0.98 (95% CI: 0.976, 0.985) for interobserver agreement
(Fig 3).
Difference in the ADC according to the type of affected
bone and clinical factors
All MRI variables including the mean, standard deviation,
and minimum and maximum ADC for the metastatic bone
tumours were analysed to assess whether they differed
according to the type of affected bone or the sex and age of
the patient; however, none of the analyses showed a strong
association. Only two of the results were statistically sig-
nificant; the standard deviation and the maximum value of
the ADC differed according to the type of affected bone
(p¼0.001 and p¼0.0213, respectively; Table 1, Fig 4). In
other words, the ADC of bone metastasis affecting irregular
bones, such as the vertebrae, sacrum, and pelvic bones,
tended to be less heterogeneous than that of long bones,
such as the humerus, femur, and tibia. In addition,
maximum ADC value tends to be lower in the metastasis
affecting irregular bones rather than long bones.
Association between the ADC value and CT radiodensity
A subanalysis was performed for the 38 patients
(M:F¼21:17; median age, 57 years) who also underwent CT
or PET/CT for the same metastatic bone tumours. Spearman
correlation analysis revealed no association between CT
radiodensity on the pre-contrast image and the mean ADC
value (p¼0.24; Fig 5). In addition, there was no significant
difference in the CT measurements including average radi-
odensity, standard deviation, and the maximum and mini-
mum value according to the five different primary cancers.
The average SUVmax in 17 patients who obtained PET/CT
was 7.13 (median, 5.8; range, 3e17.2). The analysis about the
association between SUVmax and the mean ADC value did
not show significant correlation either (p¼0.083).
Discussion
DW MRI has been widely used to study bone tumours,
injury, infection, degeneration, and other musculoskeletal
diseases, and has become an indispensable part of many
musculoskeletal MRI protocols. In cancer imaging, espe-
cially in the field of bone metastasis, it is a reliable tech-
nique for lesion detection and evaluation of treatment
response. Recently, researchers have shown that the ADC is
related to the grade or type of tumours and that DW
1120 M.J. Cha, Y.C. Yoon / Clinical Radiology 70 (2015) 1116e1121

Figure 3 (a) Intra- and (b) interobserver agreement for the mean ADC.

Figure 4 Box plot demonstrating the standard deviation and maximum value of the ADC in relation to the type of affected bone.

imaging is useful in the differentiation of cerebellar tu-
mours in children, the types and grades of some intra-axial
brain tumours, the subtypes of renal cell carcinomas, and
the histopathological grades of hepatocellular carci-
noma.10,12,14,16 In contrast, other studies have found no
Figure 5 Scatter diagram of bone metastasis segmented by mean CT
radiodensity and mean ADC.
significant association between the ADC and histopatho-
logical subtypes in pancreatic cancers and meningi-
omas.11,13 To the authors’ knowledge, however, no studies
have investigated the usefulness of DW imaging in the
characterisation of bone metastasis.
On conventional radiography and CT, bone metastases
can be classified into osteolytic, sclerotic, or mixed lesions.
It is well known that metastases from certain primary sites,
such as renal cell or thyroid carcinomas, are almost always
osteolytic, whereas those from prostatic carcinoma are
predominantly sclerotic. Metastases from breast carci-
noma, colonic carcinoma, melanoma, bladder carcinoma,
or sarcoma can also be sclerotic. In a clinical setting, the
typical appearance of certain bone metastases is often
useful in determining the type of primary malignancy.
Nonetheless, these techniques cannot provide definitive
results. Thus, the field of DW imaging would be greatly
advanced if a new MRI technique that is highly sensitive to
the cellularity and mobility of free water molecules in
tumour tissue could provide advanced information on
metastatic bone tumours; however, no significant associ-
ations were found between the ADC and the type of pri-
mary cancer for bone metastasis. Various conditions, such
as tumour cellularity, nucleus:cytoplasm ratio, size of the
tumour cells, fibrous tissues within the tumour, and
 M.J. Cha, Y.C. Yoon / Clinical Radiology 70 (2015) 1116e1121
tumoural perfusion features (i.e., micromotion of the
molecules in its capillaries) may influence the ADC value.
The results of the present study suggest that metastatic
bone tumours from five different primary cancers do not
possess different diffusion characteristics. In addition,
there does not appear to be a clear relationship between
the ADC value and CT radiodensity, which represents the
osteoblastic/osteoclastic activity of the bone tumour. This
result suggests that the hypothesis under evaluation was
too simplistic and that osteoblastic or osteoclastic activities
are just one of many features that contribute to the DW
imaging results and the ADC. No significant associations
were found between the ADC and SUVmax for bone
metastasis either, but the number of patients with PET/CT
was too small to develop useful clinical standards. It is
unclear why the ADC value of bone metastasis affecting
irregular bones was less heterogeneous with a lower
maximum value than that of long bones; however, the
results could be biased due to the small sample size, so
they should be verified in a study with a larger cohort.
The present study had several limitations. First, the
study population was relatively small, with only a small
number of patients pathologically confirmed to have me-
tastases (12 of 67). The small sample size could have been
influenced by selection and verification biases; however,
this limitation was unavoidable because patients with
bone metastases are usually in the terminal stages of their
illness, making it difficult to obtain histopathological or
clinical confirmation of the diagnosis. A larger population-
based study is required to assess the efficacy of novel MRI
techniques and DW imaging in bone metastases. Second,
the measurement of the ADC values could vary because the
ROI was drawn manually. Tiny cystic or necrotic portions
within the tumour could have been included in the ROI, as
pixel-by-pixel matching could not be performed. Further-
more, the exclusion of necrotic and cystic regions, hae-
morrhage, and bone marrow oedema was done
individually without a reference image. Nevertheless, the
intra- and interobserver agreements were excellent, which
confirms the validity of the present measurements. Third,
the ADC of each tumour was measured on only a single
axial section, so the ADC values that were measured may
not have been representative of the characteristics of the
whole tumour. Future studies in this area should consider
three-dimensional volumetric measurement of ROIs. In
addition, the novel quantitative parameters of dynamic
contrast-enhanced MRI can be studied in combination
1121
with DW imaging, as an approach to multiparametric MRI
for accurate diagnosis.
In conclusion, the present results were unable to
demonstrate any additional value of DW MRI to improve the
diagnosis of bone metastasis. DW MRI and ADC measure-
ments are not a reliable indication of the primary cancer
type in bone metastasis.
References
1. Roodman GD. Mechanisms of bone metastasis. Discov Med
2004;4:144e8.
2. Roodman GD. Mechanisms of bone metastasis. N Engl J Med
2004;350:1655e64.
3. Weatherall PT. Benign and malignant masses. MR imaging differentia-
tion. Magn Reson Imaging Clin N Am 1995;3:669e94.
4. Ma LD, Frassica FJ, McCarthy EF, et al. Benign and malignant musculo-
skeletal masses: MR imaging differentiation with rim-to-center differ-
ential enhancement ratios. Radiology 1997;202:739e44.
5. Ma LD, McCarthy EF, Bluemke DA, et al. Differentiation of benign from
malignant musculoskeletal lesions using MR imaging: pitfalls in MR
evaluation of lesions with a cystic appearance. AJR Am J Roentgenol
1998;170:1251e8.
6. Le Bihan D. Molecular diffusion nuclear magnetic resonance imaging.
Magn Reson Q 1991;7:1e30.
7. Colagrande S, Carbone SF, Carusi LM, et al. Magnetic resonance diffusion-
weighted imaging: extraneurological applications. Radiol Med
2006;111:392e419.
8. Li C, Liu ZS, Du XM, et al. Clinical value of whole-body magnetic reso-
nance diffusion weighted imaging on detection of malignant metastases.
Chin Med Sci J 2009;24:112e6.
9. Neubauer H, Evangelista L, Hassold N, et al. Diffusion-weighted MRI for
detection and differentiation of musculoskeletal tumorous and tumor-
like lesions in pediatric patients. World J Pediatr 2012;8:342e9.
10. An C, Park MS, Jeon HM, et al. Prediction of the histopathological grade
of hepatocellular carcinoma using qualitative diffusion-weighted, dy-
namic, and hepatobiliary phase MRI. Eur Radiol 2012;22:1701e8.
11. Rosenkrantz AB, Matza BW, Sabach A, et al. Pancreatic cancer: lack of
association between apparent diffusion coefficient values and adverse
pathological features. Clin Radiol 2013;68:e191e7.
12. Rumboldt Z, Camacho DL, Lake D, et al. Apparent diffusion coefficients
for differentiation of cerebellar tumors in children. AJNR Am J Neuro-
radiol 2006;27:1362e9.
13. Sanverdi SE, Ozgen B, Oguz KK, et al. Is diffusion-weighted imaging
useful in grading and differentiating histopathological subtypes of me-
ningiomas? Eur J Radiol 2012;81:2389e95.
14. Wang H, Cheng L, Zhang X, et al. Renal cell carcinoma: diffusion-
weighted MR imaging for subtype differentiation at 3.0 T. Radiology
2010;257:135e43.
15. Messiou C, Collins DJ, Morgan VA, et al. Optimising diffusion weighted
MRI for imaging metastatic and myeloma bone disease and assessing
reproducibility. Eur Radiol 2011;21:1713e8.
16. Yamasaki F, Kurisu K, Satoh K, et al. Apparent diffusion coefficient of
human brain tumors at MR imaging. Radiology 2005;235:985e91.