See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/340798445

Bone Mineral Densitometry Reporting: Pearls and Pitfalls

Article  in  Canadian Association of Radiologists Journal · April 2020

DOI: 10.1177/0846537120919627

CITATIONS READS
3 831

3 authors, including:

Patrick Martineau
BC Cancer
74 PUBLICATIONS   1,075 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Cosmological fluctuations View project

Trabecular bone score View project

All content following this page was uploaded by Patrick Martineau on 22 April 2020.

The user has requested enhancement of the downloaded file.

Musculoskeletal Radiology
Canadian Association of
Radiologists’ Journal
Bone Mineral Densitometry Reporting: 1-15
ª The Author(s) 2020
Article reuse guidelines:
Pearls and Pitfalls sagepub.com/journals-permissions
DOI: 10.1177/0846537120919627
journals.sagepub.com/home/caj

Patrick Martineau, MD, PhD1, Sarah L. Morgan, MD, MS2,

and William D. Leslie, MD, MSc3

Abstract
Dual-energy X-ray absorptiometry (DXA) is the method of choice for assessing bone mineral density (BMD). Unfortunately, the
performance and interpretation of DXA can be challenging and errors are common. In fact, it has been reported that up to 90% of BMD
reports contain at least 1 error. Errors can be the result of technique or interpretative in nature or both and can result in inappropriate
diagnosis and management. In this article, we review the various types of pitfalls frequently encountered by physicians interpreting
DXA studies. Being aware of these pitfalls will help readers recognize and avoid them when encountered in clinical practice.

Résumé
L’absorptiométrie biénergétique à rayons X (DXA) est la méthode de choix pour l’évaluation de la densité minérale osseuse
(DMO). Malheureusement, la réalisation et l’interprétation de la DXA peuvent être difficiles et les erreurs sont fréquentes. De
fait, il a été signalé que jusqu’à 90 % des rapports de DMO comportaient au moins 1 erreur. Il peut s’agir d’erreurs techniques ou
d’interprétation, ou les deux, et ces dernières peuvent entraı̂ner un diagnostic et une gestion inappropriés. Dans cet article, nous
étudions les différents types d’écueils fréquemment rencontrés par les médecins interprétant les études par DXA. Connaı̂tre ces
écueils aidera les lecteurs à les reconnaı̂tre et à les éviter quand ils les rencontreront en pratique clinique.

Keywords
osteoporosis, fractures, dual-energy X-ray absorptiometry

Introduction pharmacological management. Fenton et al2 found that nearly

two-thirds of new osteoporosis drug prescriptions were poten-
Dual-energy X-ray absorptiometry (DXA) plays an integral
tially inappropriate and often related to reporting T scores from
role in the evaluation of osteoporosis and is used for diagnosis, nonstandard sites (Ward area from the hip, lateral lumbar spine).
assessment of fracture risk, and assessment of treatment In this article, we review the various types of pitfalls that are
response. Dual-energy X-ray absorptiometry imaging exploits frequently encountered in clinical practice. Being mindful of
the differential absorption of photons of 2 different energies in these sources of pitfalls will help the interpreting physician
order to estimate the calcium-equivalent beam attenuation. recognize them when present and avoid them whenever possible.
Individual pixel-wise measures are aggregated and normalized
for the projected (2-dimensional) area and are used to deter-
mine the areal bone mineral density (BMD). A standard DXA Operator-Dependent/Technical Pitfalls
examination includes the lumbar spine, hip, and, occasionally, An important source of error in DXA scanning is under the control
the distal radius or total body, in order to determine BMD of the operator.3,6 In this section, we review those pitfalls which
values at these sites. These values are calculated through appli- may arise through the actions of the technologist overseeing the
cation of semiautomated regions-of-interest (ROIs) for the rel- study acquisition or from other technical factors.
evant bony structures, yielding absolute BMD values with
these usually expressed as T scores (young adult reference 1
population) and Z scores (age-matched reference population). Department of Radiology, University of Manitoba, Winnipeg, Manitoba,
Canada
Unfortunately, errors in DXA studies are common and can 2
Department of Medicine, University of Alabama, Birmingham, Alabama, USA
lead to incorrect patient diagnosis and management.1-5 The 3
Department of Medicine, University of Manitoba, Winnipeg, Manitoba,
majority (in one study over 90%) of DXA reports contained an Canada
error, with the majority of errors related to data analysis/interpre-
Corresponding Author:
tation, followed by incorrect patient positioning, artifacts, and William D. Leslie, MD, MSc, Department of Medicine, University of Manitoba,
demographic errors.1,4 The high DXA error rate has been shown C5121-409 Tache Avenue, Winnipeg, Manitoba, Canada R2H 2A6.
to result in a large number of patients receiving inappropriate Email: [email protected]
2 Canadian Association of Radiologists’ Journal XX(X)

Figure 1. Movement during DXA imaging can have a deleterious effect on image quality. These are 2 examples of motion artifact affecting the
hip. The left hand image is from a patient with spinal cord injury who developed severe spasms when extending the hip; the measured BMD
would be unreliable. The right hand image shows more subtle motion limited to the diaphysis below the ROI (arrow); the scan should be
repeated though BMD should not be affected. BMD indicates bone mineral density; DXA, dual-energy X-ray absorptiometry; ROI, region of
interest.

Patient Positioning Technologist retraining has been shown to be an effective

way to mitigate patient positioning errors. In a recent retrospec-
Correct patient positioning is essential in order to obtain reli-
tive study by Promma et al,3 improper patient positioning at the
able and reproducible BMD values. Proper positioning of the
spine and hip was noted in 49% and 57% of patients, respec-
patient for the posteroanterior scan of the spine requires that
tively, prior to technologist retraining. Following retraining, the
the spine be straight, the top of the iliac crests be visible, with frequencies of improper spine and hip positioning decreased to
the field of view extending from approximately the middle of approximately 9% and 13%, respectively, with the improve-
T12 to the middle of L5. For the hip scan, the technologist must ments maintained for several years following retraining.
ensure that the hip is internally rotated (with approximately
15 -25 of rotation) such that the long axis of the femoral neck
is perpendicular to the X-ray beam and that the shaft of the Patient Movement
femur is straight. Positioning can be improved through the use Patient movement can impact DXA results (Figure 1). In order
of foam blocks or positioning devices. to reduce movement-related error, it is helpful to emphasize the
A study using a spine phantom has shown that altering lor- importance of laying still and positioning patients as comfor-
dosis and kyphosis tilt angles, as well as axial and lateral sco- tably as the examination allows.
liosis, impacted on measured spinal BMD values. 7
Specifically, artifactually decreased BMD values were seen
when increasing the severity of kyphosis and lordosis, and also Impact of Pannus
when increasing the angle of axial and rotational scoliosis. The Abdominal fat can impact DXA trueness and precision, par-
degree of hip flexion has been shown to impact spinal BMD.8,9 ticularly in patients with a significant pannus overlapping
In particular, Ikegami et al8 have shown that spinal BMD was the hip region. Binkley et al16 have shown that imaging with
slightly overestimated in supine patients versus those with ele- the pannus in place or retracted could lead to unpredictable
vated legs. variation in hip BMD—up to 74%. In 49% of the men and
It has been shown that variations in femur positioning can 56% of the women, either the femoral neck, trochanter, or
affect measured hip BMD values.3,9-13 In a retrospective study total femur BMD differed by more than the facility’s mon-
involving 200 women, Rosenthall13 showed that increasing the itoring least significant change (LSC). As such, the authors
degree of internal rotation of the hip from a neutral position recommend that the pannus be retracted in routine practice
could result in either an increase (in 65% of patients) or (Figure 2).
decrease (35%) in femoral neck and total hip BMD. Several
authors have shown that hip positioning systems can help
reduce errors due to patient positioning.12,14 It should be noted Region-of-Interest Pitfalls
that radius BMD values are also prone to error in cases of Region-of-interest placement is crucial in obtaining reliable
improper positioning and/or clothing artifacts.15 BMD values. Errors in ROI placement can result in
Martineau et al 3

Figure 2. Impact of soft tissue/pannus on bone mineral density (BMD). The image on the left shows the left hip with the pannus dependent,
while the image on the right has the pannus partially retracted (tissue windows are shown below the bone windows with arrowheads delineating
the edge of the pannus). Total hip BMD decreased from 0.863 g/cm2 (T score 1.2) to 0.758 g/cm2 (T score 2.0) after pannus retraction, and
there was also a decrease in measured tissue thickness (from 28.3 to 23.0 cm) and fat percent (from 49.2% to 41.7%).

erroneous BMD values and bias serial measurements Bone mineral density is an areal density and, as such, is
(Figure 3). It has been reported that the most common source dependent on the area of the ROI used. It has been shown that
of operator-dependent error in spine BMD assessment was variances in the ROI area are commonplace and can be sources
incorrect placement of the intervertebral disk spaces for ROI of short-term bone density measurement error in both the lum-
placement.17 Fortunately, such errors rarely result in misdiag- bar spine and hip, with changes in area >2% associated with
nosis. In addition, anomalous spinal segmentation is common, significantly greater BMD measurement errors.19,20
affecting 1 in 6 cases, and can affect the ability of the operator
to identify spinal levels. The ROI misplacement errors of T12
for L1 can result in significant underestimation of BMD (mean Effect of Radionuclides and Contrast Agents
3.8% with range up to 13.1%).18 For consistency, some groups Patients undergoing DXA studies will occasionally undergo
recommend labeling from caudal to cranial. For most patients, additional imaging studies during a single visit or in close
the top of the iliac crest is at the approximate level of the L4 temporal proximity. Often, these additional imaging studies
to L5 interspace. For monitoring purposes, consistent labeling will involve administration of radioisotopes or radiological
is critical. contrast agents, which can interfere with BMD
4 Canadian Association of Radiologists’ Journal XX(X)

Figure 3. Region-of-interest placement is critical in order to obtain accurate BMD values. For the patient in (A), the edges of the lumbar spine
ROIs were incorrectly detected due to very low BMD but subsequently corrected (B). This resulted in a change of the L1 to L4 BMD value from
0.668 g/cm2 (T score 4.3) to 0.602 g/cm2 (T score 4.8). Correct identification of hip ROIs is also crucial—incorrect placement of the femoral
neck ROI (C) from failure to detect the superolateral margin of the femoral head (arrow) initially resulted in a total hip BMD of 0.945 g/cm2
(T score 0.5) which increased to 0.983 g/cm2 (T score 0.1) after correctly defining missing bone and correcting the ROI placement (bottom
right). The inferior aspect of the hip ROI extending too inferiorly in (on Hologic, it should be 10 pixels below the bottom edge of the lesser
trochanter, E), leading to a total hip BMD of 1.253 g/cm2 (T score 2.6) which, upon correction (F), decreased to 1.179 g/cm2 (T score 1.9).
Correct identification of vertebral levels is important. In patient (G), vertebral levels were initially correctly assigned giving baseline BMD
0.632 g/cm2 (T score 3.8) but were incorrectly assigned on follow-up (H) resulting in a BMD of 0.647g/cm2 (T score 3.6) suggesting no change
(difference 0.015 g/cm2). When the vertebral levels were correctly reassigned, the repeat BMD became 0.664 g/cm2 (T score 3.5) which now
exceeded the LSC indicating a significant increase (difference 0.032 g/cm2). BMD indicates bone mineral density; LSC, least significant change;
ROIs, regions of interest.

measurements as explained below. Therefore, it is important The use of iodinated contrast agents has been shown to
that standard operating procedures are in place to identify cause statistically significant but variable effects on BMD and
and delay DXA testing in order to mitigate these effects. body composition.25 No such effect was seen with gadolinium-
Fosbøl et al21 studied the effects of antecedent administra- based agents, which may be related to the much lower serum
tion of 99mTc-labeled radiotracers on patients undergoing concentration of these agents compared to iodinated contrast.25
DXA. Using both patient and phantom measurements, they The literature examining the effects of oral contrast agents on
found a significant dose-related decrease in BMD values in DXA is quite limited—case reports and clinical experience
patients who had received radiotracer injection; however, have shown that barium sulfate can artifactually increase BMD
other investigators have found no such effect. 22 Kim if overlying bone and would have the opposite effect if barium
et al,23 using a Hologic system, showed that administration is predominantly in the soft tissues adjacent to bone 26,27
of 18F-fluorodeoxyglucose also resulted in a significantly (Figure 4).
decreased whole-body BMD, decreased whole-body lean In cases where DXA will be performed along with addi-
mass, and increased whole-body fat lean body mass. The tional imaging studies, it may be beneficial to perform DXA
cause of this is most likely Compton-scattered g radiation prior to any other radiological studies in order to mitigate the
being detected by the DXA scanner. In general, Hologic potential effects of radioisotopes or contrast agents. The time
scanners are less sensitive to these effects than other for radionuclide decay is a function of the physical half-life and
scanners due to the method used for X-ray detection.24 biological clearance, and 48 hours would be sufficient for
Martineau et al 5

Figure 4. A patient received oral contrast prior to DXA imaging (left) with repeat of the study (right) once the contrast had cleared. When oral
contrast was present, the apparent L2 to L4 BMD was 1.030 g/cm2 (T score 1.4), compared to 0.846 g/cm2 (T score 3.0) without. BMD
indicates bone mineral density; DXA, dual-energy X-ray absorptiometry.

Figure 5. Varying abdominal attenuation (from ascites or significant gain/loss of abdominal fat) can make it difficult to obtain reliable lumbar
spine BMD values. The results of a DXA study in a patient before (left) and after (middle) paracentesis for significant ascites (right) are shown.
The L1 to L4 BMD value was 1.021 g/cm2 (T score 1.3) prior to drainage, increasing to 1.150 g/cm2 (T score 0.2) after drainage. BMD
indicates bone mineral density; DXA, dual-energy X-ray absorptiometry.

short-lived isotopes (18F or 99mTc). Clearance of oral barium is Interpretation Pitfalls

quite variable and may exceed a week.
The dominant source of error in DXA is interpretation
related.1,4 Fortunately, once recognized, most interpretative
pitfalls can be addressed without compromising the technical
Ascites/Loss of Abdominal Fat quality and accuracy of the study.
The presence of ascites can have a significant impact on lumbar
spine BMD values28 (Figure 5). As such, it may be beneficial to
image patients soon after paracentesis. Nonetheless, varying Vertebral Body Exclusions
amounts of ascites can render follow-up studies challenging. Exclusion of specific vertebral bodies is often necessary in
Likewise, follow-up imaging in patients with significant routine clinical practice for a number of reasons, including
weight loss (eg, following bariatric surgery) can be challenging localized degenerative changes, fractures, or other focal pathol-
due to the unpredictable effects of varying soft-tissue attenua- ogy, which may cause the BMD value obtained from a specific
tion and fat adjacent to and within the bone. vertebral level to be nonrepresentative. In a study examining
6 Canadian Association of Radiologists’ Journal XX(X)

Figure 6. Reasons to exclude vertebral levels include (A) vertebral fractures (L1), (B) severe degenerative changes (L4), (C) Paget disease (L1),
(D) bone metastases (T12, L1, L3), (E) laminectomy (L3, L4), (F) spinal fusion (L4), and (G) vertebroplasty (L3).

over 22 000 clinical cases, Tsang and Leslie29 reported that for the exclusion of vertebral bodies has been shown to be only
vertebral body exclusions were present in approximately 30% moderate.30 Hansen et al31 have shown that use of an atlas
of cases. Vertebral body exclusion was associated with a small could significantly improve interobserver agreement for the
but significant improvement in fracture risk prediction. Verteb- exclusion of specific vertebral levels (Figure 6).
ral exclusion should not be based on numeric criteria alone. It is
important that the interpreting physicians review the DXA
images and ancillary imaging, attempt to identify the cause for Nonosseous Causes of BMD Errors
the anomalous BMD value, and subsequently decide whether The interpreting physician must be vigilant for a number of
the particular vertebral level should be excluded, since artifacts nonosseous factors that can lead to errors in BMD. In particu-
produce discordantly higher or lower BMD. Vertebral body lar, cholelithiasis, nephrolithiasis, and calcified mesenteric
exclusion is at the discretion of the interpreting physician, lymph nodes are commonly seen on spinal DXA studies (Fig-
which introduces a degree of ambiguity when assessing spinal ure 7). At the hip, heterotopic ossification is frequently seen. If
BMD. Interobserver agreement among interpreting physicians not addressed, these findings can have an impact on BMD,
Martineau et al 7

Figure 7. Nonosseous calcifications are frequently encountered on dual-energy X-ray absorptiometry studies. Some examples are (A) calcified
gallstones, (B) calcified mesenteric nodes, (C) staghorn renal calculus, (D) severe aortic calcification, and (E) heterotopic ossification.

even when the abnormality is completely confined to the soft Abdominal aortic calcification (AAC) is frequently seen
tissues. When included within bone ROIs, these densities will on DXA studies and will often overly the lumbar spine on the
directly increase the calculated BMD. Calcific densities lateral posteroanterior DXA images. The presence of aortic calcifi-
to the spine ROIs will lower vertebral BMD by overestimating cations can easily be confirmed on the lateral view or ver-
the density of the adjacent soft tissue, leading to underestima- tebral fracture analysis (VFA) scan. The results of studies
tion.32 In addition to calcifications, metallic densities, either examining the effect of AAC on lumbar spine BMD have
internal or external to the patient, are frequent findings on DXA been conflicting. Some studies have shown that AAC has no
studies and include bra wires, jewelry, spinal fusion hardware, significant effect on BMD,37-39 while others have reported a
and other medical devices that can artifactually increase (and small effect.40-42 Bristow et al41 have suggested that, on
occasionally decrease) BMD33 (Figure 8). The effects of extra- follow-up studies, the combination of increasing AAC and
neous densities lateral to the spine can be mitigated through the decreasing vertebral body areas can incorrectly suggest stable
use of software corrections (Figure 9). spinal BMD values, despite an overall decrease in bone
On Hologic scanners, dense metal such as a bullet or tanta- mineral content.
lum clips overlapping the bone can cause a ‘‘black hole arti-
fact,’’ which can decrease the measured BMD value 34
(Figure 8E). This may be caused by the extreme densities Osseous Causes of Focally Increased BMD
encountered producing absorption far beyond the range typical There are a number of pathologies that can result in focally
of bone and tissue such that the difference between the low- increased BMD. Almost invariably, these increased values are
and high-energy beams is unpredictable. spurious and must be recognized as such by the interpreting
Patient clothing can also have an impact on BMD15,35,36 physician. An example of this is Paget disease that can affect
(Figure 10). Phantom studies haves shown that thick or reflec- the spine, pelvis, or proximal femurs and accounts for approx-
tive clothing can negatively impact DXA precision.35,36 This imately 1.4% of abnormally elevated BMD values.43 Radiolo-
can be mitigated by having the patient avoid wearing clothing gically, Paget is usually recognized by bony expansion, cortical
made of dense materials (such as wool or denim), reflective thickening, and coarsening of the trabeculae, which generally
material, or that which uses metallic thread. appears as diffuse sclerosis on DXA images (Figure 11). Bone
8 Canadian Association of Radiologists’ Journal XX(X)

Figure 8. Metallic artifacts are commonly seen on dual-energy X-ray absorptiometry images. A, Aortic endovascular stent, (B) insulin pump, (C)
left ventricular assist device with leads, (D) mesh from inguinal hernia repair, (E) bullets (note the presence of the ‘‘black hole’’ artifacts), (F)
spinal fusion hardware, (G) lap band port, and (H) surgical staples.

mineral density values obtained from sites affected by Paget changes in the spine have a larger impact on BMD than those
disease should be omitted. in the hip. In either case, recognition of the presence of degen-
Sclerotic metastases, arising either from prostate or breast erative changes is usually straightforward, and the interpreting
cancer, are not rare in patients undergoing DXA who may physician must decide on the appropriateness of using the
undergo serial BMD monitoring during androgen deprivation affected site for BMD evaluation. In patients with extensive
therapy and aromatase inhibitor therapy, respectively. These osteoarthritic changes, imaging the one-third radius site may be
are generally easily recognized on DXA images as areas of beneficial due to the lack of degenerative changes at this site.
focal sclerosis. Occasionally, benign bone neoplasms (such Compression fractures of the spine are a frequent finding in
as enostoses or osteoblastomas) or bone infarcts may present patients undergoing BMD evaluation. The importance of
similarly (Figure 12). Likewise, BMD values obtained from recognizing these fractures is 2-fold: The presence of such a
sites affected by bone lesions should be omitted, and it may fracture is an independent factor for risk fracture assessment,
be indicated to recommend further investigations. and fractured vertebrae should be excluded from the analysis as
Osteoarthritic changes are almost universal phenomena in these will artifactually increase the calculated BMD. Fortu-
older patients and are the most commonly encountered source nately, these are usually easily recognized due to the loss of
of artifact on DXA images. In the spine, these manifest as end- vertebral body height and the sclerotic appearance. When in
plate osteophytosis, localized or diffuse sclerosis, loss of disk doubt, review of previous spine imaging or VFA can clarify the
height, and facet joint arthropathy, while in the hip, these are situation.
seen as joint space narrowing and subchondral sclerosis, with
‘‘buttressing’’ of the medial femoral neck (calcar). Osteoar-
thritic changes in the spine are usually multilevel but are fre- Osseous Causes of Focally Decreased BMD
quently more marked at 1 or more level. Osteoarthritic changes Patients with spinal stenosis are frequently treated with lami-
at both the spine and the hip will artifactually increase BMD at nectomy of 1 or more lumbar levels. The resultant absence of
these sites.37,44-51 Liu et al44 have shown that osteoarthritic the laminae and spinous processes reduces the measured BMD
Martineau et al 9

Figure 9. Extraneous densities on DXA studies can, in some cases, be automatically detected and subtracted by the software. In this case (A),
the abdominal pacemaker and wires lateral to the spine were detected (delineated in yellow) and omitted from the soft-tissue baseline; however,
it is important that the interpreting physician review the images to ensure that metallic densities are correctly identified by the system. In this
case, L2 was omitted because of overlying artifacts. Calcific densities can also be removed by the software. In (B), a renal stone was initially
included in the background, leading to a L3 BMD of 0.875 g/cm2 (T score 1.9). After removing it from the background, the BMD became
0.850 g/cm2 (T score 2.1). BMD indicates bone mineral density; DXA, dual-energy X-ray absorptiometry.

values at the corresponding vertebral levels. Fortunately, lami- these conditions are generally long-standing and may be indicated
nectomy defects are usually easily recognized on lumbar spine on the requisition form or the patient questionnaire.
DXA images. Spina bifida, associated with congenital absence More commonly encountered, rheumatological conditions
of the posterior vertebral elements, can present with the same such as ankylosing spondylitis (AS) and diffuse idiopathic ske-
findings. In both cases, omission of the affected vertebral levels letal hyperostosis (DISH) can show increased BMD throughout
is indicated. the lumbar spine.
Lytic osseous lesions can affect either the vertebra or hip Paradoxically, despite the increased BMD values seen in these
and are usually metastatic in etiology. As these are occasionally conditions, affected patients are known to be at increased risk of
first seen on DXA images, it is important to recognize these due fractures. The reason for this is easily understood—both condi-
to the potential clinical implications. Benign lucent lesions, tions manifest with ossification of spinal ligaments (anterior and
including hemangiomas, bony cysts, and fibrous dysplasia, are posterior spinal ligaments in DISH, and interspinous ligaments in
all potential causes of focally decreased BMD values. As in the AS) but decreased trabecular bone density. In terms of the appen-
case of sclerotic lesions, omission of the affected sites is usu- dicular skeleton, DISH can be associated with increased BMD
ally indicated, and additional imaging to further characterize values in the hip and distal radius, while AS is generally associ-
the lesion can be helpful. ated with decreased BMD values at those sites (Figure 13).
A commonly encountered cause of diffusely increased
BMD is diffuse sclerotic osseous metastases, as can be seen
Osseous Causes of Diffusely Increased BMD in cases of prostate or breast cancer (Figure 14).
There are numerous systematic causes of diffusely increased
BMD. These include a number of heritable conditions (eg, osteo-
petrosis, pyknodysostosis), metabolic disorders (eg, hypervitami-
Bone Mineral Density for Monitoring Therapy
nosis A and D, fluorosis, renal osteodystrophy, hepatitis C– One of DXA’s great strengths is the ability to detect small
associated osteosclerosis), as well as a number of other conditions changes in BMD on consecutive examinations, a key aspect in
(eg, mastocytosis, hypertrophic osteoarthropathy) but, overall, assessing trends and treatment response. Evidence from clin-
diffusely increased BMD is uncommonly seen.52 Furthermore, ical trials and clinical registries shows that treatment-related
10 Canadian Association of Radiologists’ Journal XX(X)

Figure 10. Occasionally, patient clothing can have an impact on BMD. A, This patient wore flame-retardant pants, which were clearly delineated
on a total-body DXA study (B). A more subtle finding is seen in (C) where a waistband is seen overlying L3, which lead to the image being
repeated (D). Also included in (C) and (D) are bra fasteners, which are not included within the ROIs. BMD indicates bone mineral density; DXA,
dual-energy X-ray absorptiometry; ROI, region of interest.

Figure 11. Paget disease is frequently encountered on DXA. Paget must be recognized as it can have an important effect on BMD values. This
patient had Paget involving the left femur (left) leading to a total hip BMD of 1.664 g/cm2 (T score 5.2). As a result, the right femur was imaged
(right), yielding a total hip BMD of 0.852 g/cm2 (T score 1.2). BMD indicates bone mineral density; DXA, dual-energy X-ray absorptiometry.
Martineau et al 11

Figure 12. Benign bone lesions are frequently encountered on dual-energy X-ray absorptiometry. In (A), a calcified enchondroma is present in
the left proximal femur. Likewise, a bone infarct (arrow) is present in (B).

Figure 13. Appearance of ankylosing spondylitis on DXA with ossification of the spinal ligaments (particularly the interspinous ligaments). The
L1 to L4 BMD was 1.224 g/cm2 (T score 0.4) while the total hip BMD was 0.797 g/cm2 (T score 1.7). BMD indicates bone mineral density; DXA,
dual-energy X-ray absorptiometry.

increases in BMD are associated with reduced fracture risk the LSC is crucial so that, when reporting differences on serial
compared with stable BMD, whereas decreases in BMD are BMD measurements, only those changes that meet or exceed
associated with higher risk for fractures53,54; however, it is the LSC are considered statistically significant. However, a
important to realize that despite increases in BMD, patients statistically significant change in BMD exceeding the LSC
can still have ongoing fractures. The 95% least significant may not always be clinically significant. Procedures for deter-
change (LSC) is the smallest change that can be reliably mining the LSC are well described.55-57 Problems can arise
detected (with 95% confidence). Accurate determination of when the sample size used to calculate the LSC is not
12 Canadian Association of Radiologists’ Journal XX(X)

Figure 14. Typical appearance of diffuse bone metastases from breast cancer on DXA (left) with correlative CT (right). Note the diffusely
sclerotic appearance of all vertebral bodies and iliac wings, and the markedly elevated BMD values. The L1 to L4 and total hip BMDs were 1.693
(T score 4.3) and 1.149 g/cm2 (T score 1.1), respectively. For reference, the one-third radius BMD was 0.641 g/cm2 (T score 2.7). BMD
indicates bone mineral density; CT, computed tomography; DXA, dual-energy X-ray absorptiometry.

Figure 15. Atypical femoral fractures (AFFs) can be subtle, particularly on DXA. In this patient, an area of sclerosis is visible along the lateral
aspect of the right proximal femoral diaphysis on DXA (left) with an apparent lucency (arrow) confirmed on the subsequent plain film. DXA
indicates dual-energy X-ray absorptiometry.

sufficiently large58 or when percent change (not absolute nonvertebral fracture risk reduction.54 Overall, how best to
change) is used.59 The most commonly used method for esti- measure and report changes in BMD, relationship with frac-
mating test–retest measurement error (same-day, same- ture risk and optimal BMD testing interval is an evolving
technologist, simple repositioning) will tend to underestimate subject and clinicians should be cautious when attributing
long-term measurement error (different days and technolo- significance to small changes seen on serial DXA
gists).60 Paradoxically, although treatment-related increases examinations.
in spine BMD exceed those seen at the hip, due to age-
related degenerative changes observed changes in spine BMD
Atypical Femoral Fractures/Postdenosumab-Rebound
are actually less strongly associated with fracture risk than
changes in total hip BMD.61 It is important to note that BMD Vertebral Fractures
change accounts for only half of the observed reduction in Atypical femoral fractures (AFFs) are a type of insufficiency
vertebral fracture risk and a smaller proportion of fracture linked to prolonged use of bisphosphonates and the
Martineau et al 13

RANKL inhibitor denosumab, which can be encountered in 5. Cetin A, Ozgüçlü E, Ozçakar L, Akinci A. Evaluation of the
patients undergoing DXA. Specific diagnostic criteria for AFFs patient positioning during DXA measurements in daily clinical
have been proposed by the American Society of Bone and practice. Clin Rheumatol. 2008;27(6):713-715.
Mineral Research,62 and the radiographic findings have previ- 6. Fuleihan GE-H, Testa MA, Angell JE, Porrino N, Leboff MS.
ously been reviewed.63 It is important that the interpreting Reproducibility of DXA absorptiometry: a model for bone loss
physician review all DXA images of the hip for signs of peri- estimates. J Bone Miner Res. 1995;10:1004-1014.
osteal thickening or incomplete fracture along the lateral aspect 7. Izadyar S, Golbarg S, Takavar A, Zakariaee SS. The effect of the
of the femoral diaphysis (Figure 15). Bilateral full femur DXA lumbar vertebral malpositioning on bone mineral density mea-
images can be used to detect AFFs.64 surements of the lumbar spine by dual-energy X-ray absorptio-
In contrast to bisphosphonates that have a long skeletal metry. J Clin Densitom. 2016;19:277-281.
residence time and where a ‘‘drug holiday’’ can be considered 8. Ikegami S, Kamimura M, Uchiyama S, Nakamura Y, Mukaiyama
after 3 to 5 years of treatment, patients who have recently K, Kato H. Clinical implications of hip flexion in the measure-
discontinued denosumab experience a rapid loss in effect with ment of spinal bone mineral density. J Clin Densitom. 2016;19(1):
accelerated bone turnover and risk of rebound vertebral frac- 270-276.
tures. There are most commonly seen at T12 and L1 and may 9. Lekamwasam S, Sumith R, Lenora J. Effect of leg rotation on hip
be multiple, occurring 8 to 16 months following the last deno- bone mineral density measurements. J Clin Densitom. 2003;
sumab injection, and often in patients with previous vertebral 6(10):331-336.
fracture.65,66 Recognition of these fractures is important as it 10. Çelik Ö, Salcı Y, Manisalı M, Korkusuz F. The effect of hip
has been shown that vertebroplasty in these patients is associ- rotation on bone mineral density of the proximal femur measured
ated with additional vertebral fractures.65,66 The physician by dual energy X-ray absorptiometry. Measurement. 2009;2(3):4.
interpreting DXA, who usually has access to the patient’s treat- 11. Cheng XG, Nicholson PHF, Boonen S, et al. Effects of antever-
ment history and DXA studies, is well positioned to raise the sion on femoral bone mineral density and geometry measured by
possibility of rebound vertebral fractures. dual energy X-ray absorptiometry: a cadaver study. Bone. 1997;
21(5):113-117.
12. Goh JCH, Low SL, Bose K. Effect of femoral rotation on bone
Conclusion
mineral density measurements with dual energy X-ray absorptio-
Dual-energy X-ray absorptiometry is a powerful and useful metry. Calcif Tissue Int. 1995;57(1):340-343.
quantitative imaging modality but—perhaps even more than 13. Rosenthall L. Range of change of measured BMD in the femoral
other imaging studies—close attention must be paid to all neck and total hip with rotation in women. J Bone Miner Metab.
aspects of the study acquisition and interpretation in order to 2004;22(8):496-499.
avoid erroneous and misleading results. 14. Hans D, Duboeuf F, Schott AM, et al. Effects of a new positioner
on the precision of hip bone mineral density measurements.
Declaration of Conflicting Interests J Bone Miner Res. 1997;12(4):1289-1294.
The author(s) declared the following potential conflicts of interest 15. Krueger D, Vallarta-Ast N, Libber J, Checovich M, Gangnon R,
with respect to the research, authorship, and/or publication of this Binkley N. Positioner and clothing artifact can affect one-third
article: S.L.M. has served as a consultant for Amgen. radius BMD measurement. J Clin Densitom. 2013;16(2):154-159.
16. Binkley N, Krueger D, Vallarta-Ast N. An overlying fat pannicu-
Funding lus affects femur bone mass measurement. J Clin Densitom. 2003;
The author(s) received no financial support for the research, author- 6(6):199-204.
ship, and/or publication of this article. 17. Staron RB, Greenspan R, Miller TT, Bilezikian JP, Shane E,
Haramati N. Computerized bone densitometric analysis:
References operator-dependent errors. Radiology. 1999;211(4):467-470.
1. Krueger D, Shives E, Siglinsky E, et al. DXA errors are common 18. Peel NFA, Johnson A, Barrington NA, Smith TWD, Eastell R.
and reduced by use of a reporting template. J Clin Densitom. Impact of anomalous vertebral segmentation on measurements of
2019;22(2):115-124. bone mineral density. J Bone Miner Res. 1993;8(2):719-723.
2. Fenton JJ, Robbins JA, Amarnath ALD, Franks P. Osteoporosis 19. Leslie WD. The impact of bone area on short-term bone density
overtreatment in a regional health care system. JAMA Intern Med. precision. J Clin Densitom. 2006;9(3):150-153.
2016;176(3):391-393. 20. Kline GA, Hanley DA. Differences of vertebral area in serial bone
3. Promma S, Sritara C, Wipuchwongsakorn S, et al. Errors in density measurements: a common source of potential error in
patient positioning for bone mineral density assessment by dual interpretation of BMD change. J Clin Densitom. 2006;9(2):
X-ray absorptiometry: effect of technologist retraining. J Clin 419-424.
Densitom. 2018;21(2):252-259. 21. Fosbøl MØ, Dupont A, Alslev L, Zerahn B. The effect of 99mTc
4. Messina C, Bandirali M, Sconfienza LM, et al. Prevalence and on dual-energy X-ray absorptiometry measurement of body com-
type of errors in dual-energy X-ray absorptiometry. Eur Radiol. position and bone mineral density. J Clin Densitom. 2013;16(3):
2015;25(1):1504-1511. 297-301.
14 Canadian Association of Radiologists’ Journal XX(X)

22. Xie LJ, Li JF, Zeng FW, Jiang H, Cheng MH, Chen Y. Is bone 38. Frye MA, Melton LJ, Bryant SC, et al. Osteoporosis and calcifi-
mineral density measurement using dual-energy X-ray absorptio- cation of the aorta. Bone Miner. 1992;19(2):185-194.
metry affected by gamma rays? J Clin Densitom. 2013;16(5): 39. Reid IR, Evans MC, Ames R, Wattie DJ. The influence of osteo-
275-278. phytes and aortic calcification on spinal mineral density in post-
23. Kim D-W, Kim WH, Kim MH, et al. Effect of 18F-FDG admin- menopausal women. J Clin Endocrinol Metab. 1991;72(6):
istration on measurements of bone mineral density and body com- 1372-1374.
position by dual-energy X-ray absorptiometry. J Clin Densitom. 40. Banks LM, Lees B, Macsweeney JE, Stevenson JC. Effect of
2013;16(1):283-286. degenerative spinal and aortic calcification on bone density mea-
24. Gumuser G, Parlak Y, Topal G, Aras F, Ruksen E, Sayit E. Effects surements in post-menopausal women: links between osteoporo-
of 99mTc-MIBI and 99mTc-MDP administration on dual-energy sis and cardiovascular disease? Eur J Clin Invest. 1994;24(4):
X-ray absorptiometry bone mineral density measurements. Nucl 813-817.
Med Commun. 2009;30(2):445-448. 41. Bristow SM, Gamble GD, Horne AM, Reid IR. Longitudinal
25. Sala A, Webber C, Halton J, et al. Effect of diagnostic radio- changes in bone mineral density, bone mineral content and bone
isotopes and radiographic contrast media on measurements of area at the lumbar spine and hip in postmenopausal women, and
lumbar spine bone mineral density and body composition by the influence of abdominal aortic calcification. Bone Rep. 2019;
dual-energy X-ray absorptiometry. J Clin Densitom. 2006;9(3): 10(6):100190.
91-96. 42. Orwoll ES, Oviatt SK, Mann T. The impact of osteophytic and
26. Suna Kiraç F, Yüksel D, Taskaya Yaylali O. Pitfalls in the mea- vascular calcifications on vertebral mineral density measurements
surement of bone mineral density by the dual-energy X-ray in men. J Clin Endocrinol Metab. 1990;70(5):1202-1207.
absorptiometric method. Clin Nucl Med. 2001;26:874. 43. Gregson CL, Steel SA, O’Rourke KP, et al. ‘Sink or swim’: an
27. Spencer RP, Malcolm DM, Barton PA. Totem pole sign. Bone
evaluation of the clinical characteristics of individuals with high
densitometry study with retained barium. Clin Nucl Med. 1991;
bone mass. Osteoporos Int. 2012;23(1):643-654.
16:596.
44. Liu G, Peacock M, Eilam O, Dorulla G, Braunstein E, Johnston
28. Mann ML, Thornley-Brown D, Campbell R, et al. The effect of
CC. Effect of osteoarthritis in the lumbar spine and hip on bone
peritoneal dialysate on DXA bone densitometry results in patients
mineral density and diagnosis of osteoporosis in elderly men and
with end-stage renal disease. J Clin Densitom. 2008;11(4):
women. Osteoporos Int. 1997;7(3):564-569.
532-536.
45. Pye SR, Reid DM, Adams JE, Silman AJ, O’Neill TW. Radio-
29. Tsang JF, Leslie WD. Exclusion of focal vertebral artifacts from
graphic features of lumbar disc degeneration and bone mineral
spine bone densitometry and fracture prediction: a comparison of
density in men and women. Ann Rheum Dis. 2006;65(3):
expert physicians, three computer algorithms, and the minimum
234-238.
vertebra. J Bone Miner Res. 2007;22(1):789-798.
46. Ichchou L, Allali F, Rostom S, et al. Relationship between spine
30. Hansen KE, Binkley N, Christian R, et al. Interobserver reprodu-
osteoarthritis, bone mineral density and bone turn over markers in
cibility of criteria for vertebral body exclusion. J Bone Miner Res.
post menopausal women. BMC Womens Health. 2010;10:25.
2005;20(3):501-508.
47. Ito M, Hayashi K, Yamada M, Uetani M, Nakamura T. Relation-
31. Hansen KE, Binkley N, Blank RD, et al. An atlas improves inter-
observer agreement regarding application of the ISCD vertebral ship of osteophytes to bone mineral density and spinal fracture in
body exclusion criteria. J Clin Densitom. 2007;10(3):359-364. men. Radiology. 1993;189(4):497-502.
32. Morgan SL, Lopez-Ben R, Nunnally N, et al. The effect of com- 48. Orwoll ES, Oviatt SK, Mann T. The impact of osteophytic and
mon artifacts lateral to the spine on bone mineral density in the vascular calcifications on vertebral mineral density measurements
lumbar spine. J Clin Densitom. 2008;11(2):243-249. in men. J Clin Endocrinol Metab. 1990;70(2):1202-1207.
33. Martineau P, Bazarjani S, Zuckier LS. Artifacts and incidental 49. Yu W, Glüer CC, Fuerst T, et al. Influence of degenerative joint
findings encountered on dual-energy X-ray absorptiometry: atlas disease on spinal bone mineral measurements in postmenopausal
and analysis. In: Bayazi K, ed. Seminars in Nuclear Medicine. women. Calcif Tissue Int. 1995;57(3):169-174.
Amsterdam, the Netherlands: Elsevier; 2015:458-469. 50. Rand TH, Seidl G, Kainberger F, et al. Impact of spinal degen-
34. Morgan SL, Lopez-Ben R, Nunnally N, et al. Black hole arti- erative changes on the evaluation of bone mineral density with
facts—a new potential pitfall for DXA accuracy? J Clin Densi- dual energy X-ray absorptiometry (DXA). Calcif Tissue Int. 1997;
tom. 2008;11(2):266-275. 60(5):430-433.
35. McNamara EA, Feldman AZ, Malabanan AO, et al. Effect of 51. Rand T, Schneider B, Grampp S, Wunderbaldinger P, Migsits H,
clothing on measurement of bone mineral density. J Clin Densi- Imhof H. Influence of osteophytic size on bone mineral density
tom. 2016;19(3):216-219. measured by dual X-ray absorptiometry. Acta Radiol. 1997;38(5):
36. Siglinsky E, Binkley N, Krueger D. Do textiles impact DXA bone 210-213.
density or body composition results? J Clin Densitom. 2018; 52. Whyte MP.Misinterpretation of osteodensitometry with high bone
21(8):303-307. density: BMD Z  þ 2.5 is not ‘‘normal.’’ J Clin Densitom. 2005;
37. Drinka PJ, DeSmet AA, Bauwens SF, Rogot A. The effect of 8(1):1-6.
overlying calcification on lumbar bone densitometry. Calcif Tis- 53. Leslie WD, Majumdar SR, Morin SN, Lix LM. Change in bone
sue Int. 1992;50(1):507-510. mineral density is an indicator of treatment-related antifracture
Martineau et al 15

effect in routine clinical practice: a registry-based cohort study. 60. Leslie WD. Factors affecting short-term bone density precision
Ann Intern Med. 2016;165(1):465-472. assessment and the effect on patient monitoring. J Bone Miner
54. Bouxsein ML, Eastell R, Lui LY, et al. Change in bone density Res. 2008;23(4):199-204.
and reduction in fracture risk: a meta-regression of published 61. Leslie WD, Martineau P, Bryanton M, Lix LM. Which is the
trials. J Bone Miner Res. 2019;34(4):632-642. preferred site for bone mineral density monitoring as an indicator
55. Jankowski LG, Warner S, Gaither K, et al. Cross-calibration, of treatment-related anti-fracture effect in routine clinical prac-
least significant change and quality assurance in multiple tice? A registry-based cohort study. Osteoporos Int J. 2019;30(6):
dual-energy X-ray absorptiometry scanner environments: 1445-1453.
2019 ISCD official position. J Clin Densitom. 2019;22(4): 62. Shane E, Burr D, Abrahamsen B, et al. Atypical subtrochanteric
472-483. and diaphyseal femoral fractures: second report of a task force of
56. Kendler DL, Compston J, Carey JJ, Wu CH, Ibrahim A, the American Society for Bone and Mineral Research. J Bone
Lewiecki EM. Repeating measurement of bone mineral Miner Res. 2014;29(3):1-23.
density when monitoring with dual-energy X-ray absorptiome- 63. Khan AA, Leslie WD, Lentle B, et al. Atypical femoral fractures:
try: 2019 ISCD official position. J Clin Densitom. 2019;22(1): a teaching perspective. Can Assoc Radiol J. 2015;66(2):102-107.
489-500. 64. Cheung AM, McKenna MJ, van de Laarschot DM, et al.
57. Siminoski K, O’Keeffe M, Lévesque J, Hanley D, Brown JP. Detection of atypical femur fractures. J Clin Densitom. 2019;
Canadian Association of Radiologists technical standards for 22(4):506-516.
bone mineral densitometry reporting. Can Assoc Radiol J. 65. Anastasilakis AD, Polyzos SA, Makras P, Aubry-Rozier B,
2013;64(3):281-294. Kaouri S, Lamy O. Clinical features of 24 patients with
58. Leslie WD, Moayyeri A; Manitoba Bone Density Program. Min- rebound-associated vertebral fractures after denosumab disconti-
imum sample size requirements for bone density precision assess- nuation: systematic review and additional cases. J Bone Miner
ment produce inconsistency in clinical monitoring. Osteoporos Res. 2017;32(6):1291-1296.
Int J. 2006;17(11):1673-1680. 66. Lamy O, Gonzalez-Rodriguez E, Stoll D, Hans D, Aubry-Rozier
59. Leslie WD; Manitoba Bone Density Program. The importance of B. Severe rebound-associated vertebral fractures after denosumab
spectrum bias on bone density monitoring in clinical practice. discontinuation: 9 clinical cases report. J Clin Endocrinol Metab.
Bone. 2006;39(1):361-368. 2017;10(2):354-358.

View publication stats