Painful bone marrow edema of the knee: differential
diagnosis and therapeutic concepts
Siegfried Hofmann, MD
, Josef Kramer, MD
, Anosheh Vakil-Adli, MD
Nicolas Aigner, MD
, Martin Breitenseher, MD
General and Orthopaedic Hospital Stolzalpe, 8852 Stolzalpe, Austria
Institute MR and CT Diagnostic at the Schillerpark, 4020 Linz, Austria
St. Vincents Hospital, 4020 Linz, Austria
Orthopaedic Hospital, Speising, Vienna, Austria
Department of Osteology, University Radiology Clinic, Vienna, Austria
During the last decade, MRI has been the imaging
modality of choice for evaluating patients with pain-
ful bones or joints with normal or unspecific radio-
graphs . Characteristic MRI signal alterations have
been described for most bone and joint diseases, and
MRI examination has become the diagnostic standard
for many patients with painful knee joints. The most
important functional units of the joint, in particular
cartilage, subchondral bone, the capsular–ligament
complex, and the surrounding soft tissues, can be
visualized simultaneously with MRI  . For thera-
peutic decision making, the correct interpretation of
the MRI findings is of utmost importance . Bone
marrow edema (BME), with its typical signal appear-
ance on MRI, is a common but nonspecific signal
pattern that can be found in the bony parts of the
joints in several diseases [4,5]. Bone marrow edema
can be categorized into three distinct groups accord-
ing to cause:
1. Ischemic BME
Bone marrow edema syndrome (BMES)
Osteochondritis dissecans (OCD)
Complex regional pain syndrome (CRPS)
2. Mechanical BME
Bone contusion (bone bruise)
3. Reactive BME
Because only marrow structures are involved in
BME, plain-film radiographs and CT are unable to
detect changes with sufficient sensitivity. Bone scin-
tigraphy can detect early changes in vascularization
in areas with BME by increased tracer accumulation
, but its spatial anatomic resolution is poor, and
differentiation from other disorders characterized by
increased tracer uptake is generally impossible .
BME is not visible with arthroscopy. Only MR
imaging provides adequate detection of BME. BME
is characterized by low signal intensity compared
with unaffected bone marrow on T1-weighted
images. O n T2-weighted images, especially when
fat-suppression techniques are used, high signal in-
tensities in the low-signal areas of the T1-weighted
images are typical for BME (Fig. 1). After intrave-
nous administration of contrast agents, enhancement
of the BME lesion is more evident, indicating hyper-
vascularity and repair activity . On histologic
examination, BME is caused by increased intra- and
0030-5898/04/$ – see front matter D 2004 Elsevier Inc. All rights reserved.
* Corresponding author.
E-mail address: firstname.lastname@example.org
Orthop Clin N Am 35 (2004) 321–333
extracellular fluid within bone marrow inducing new
bone formation and repair processes .
Pain is caused by the increased intraosseous
pressure (normal pressusre, 20–30 mm Hg) because
of the abnormally high fluid content in the marrow
spaces [10–12]. The characteristic symptom of BME
in the knee is pain during mechanical loading com-
bined with more or less severe complaints during
night. Also typical is pain in the affected area when it
is tapped . Until now there has been no explana-
tion why the intensity or extent of BME in MRI does
not always correlate with pain. Sometimes BME of
the knee is observed in asymptomatic patients ;
on the other hand, BM E is som etimes th e only
definite morphologic alteration in long- lasting com-
plaints . This article discusses the various causes of
painful BME of the knee joint. The aim is to allow
proper diagnosis using clinical, radiographic and
MRI findings. The therapeutic concepts for the dif-
ferent BME entities are also addressed.
Ischemic bone marrow edema
A common cause of BME is an ischemic process,
which frequently is combined with other causative
factors. Ischemic BME includes osteonecrosis, OCD,
Osteonecrosis is characterized by ischemic necro-
sis of bony structures (bone marrow, trabecula e,
cortex) in the epiphysis of convex joint compart-
ments. By far the most frequently involved location
is the hip joint, followed by the knee. In osteonecrosis
Fig. 1. Bone marrow edema syndrome (BMES): typical signal pattern of diffuse BME in different MRI sequences of coronal
images. (A) T1-weighted image. (B) Fat-suppressed T2-weighted image. (C) Fat suppressed T2-weighted image after contrast
media. Note the diffuse extension of one complete quadrant of the knee.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333322
of the knee joint, two distinct types, secondary and
spontaneous osteonecrosis [SONK]) can be differenti-
ated [14,15]. Secondary osteonecrosis can be observed
in younger patients ( 20 – 55 years) who frequently
have typical risk factors for osteonecrosis. Multiple
necrosis and infarcts involving both knee joints and
other sites can be detected in most of these cases. The
etiology of secondary osteonecrosis is similar to that
of osteonecrosi s of th e hip for s ever al ische mic
factors . In addition, the course of the disease
can be staged like osteonecrosis of the hip joint .
Initially, in the reversible initial stage (ARCO I), there
are only focal subchondral areas of BME. Frequently,
however, the location is not confined to the loading
zone. At this time the BME pattern is nonspecific.
In irreversible early stage (ARCO II), a subchondral
osteonecrotic area is surrounded by a reactive inter-
face (Fig. 2). In this stage, plain radiographs are still
negative and therefore are not useful for diagnostic
evaluation. Concomitant BME is commonly seen ad-
jacent to the necrotic area. This appearance may en-
large the necrotic area on T2-weighted images or even
hide a small subchondral necrotic lesion in the early
phase. Diagnosis of secondary osteo necrosis with
concomitant edema can be made without difficulty
in most cases, however, by the pathognomonic signal
changes on MR [15,17].
SONK is observed in older patients ( > 55 years)
without the classic osteonecrosis risk factors. There is
a predominance of females with isolated involvement
of the medial femoral condyle [15,16]. In a prospec-
tive MR study of 176 patients suffering from knee
pain, 3.4% had SONK. For patients over 65 years of
age, the prevalence was even higher (9.4%) . MRI
and histology demonstrate subchondral microfrac-
tures combined with ischemic necrosis [19,20]. The
course of SONK can be also divided into different
stages, but the appearance and course differ signifi-
cantly from second ary osteonecrosis . Initially
(stage I), subchondral BME is observed in the load-
bearing zone of the femoral condyle (Fig. 3). Involve-
ment of the weight-bearing zone of the tibia is
uncommon . Normally, because of this typical
location and the age of the patient, a definite diagno-
sis can be made in this early phase. The initial stage is
reversible. Further progression leads to early sub-
chondral fracture with flattening of the condyle (stage
II), to osteochondral fracture (stage III) and, subse-
quently, to secondary osteoarthritis (stage IV) .
Bone marrow edema syndrome
Whether BMES represents a distinct disease (tran-
sient osteoporosis, algodystrophy, transient bone mar-
row edema syndrome) [21,22] or is a subtype of
osteonecrosis [23–25] remains controversial. Most
likely, BMES is caused by diffuse subacute ischemia,
which completely heals in most cases because of a
sufficient repair mechanism . No pathologic plain
radiographic findings can be observed during the
initial period of 4 to 6 weeks. Afterwards, a slight
demineralization of the affected area can be seen. As
a result, BMES is often also referred to as transient
osteoporosis. Histologic examination shows no signs
characteristic of osteoporosis, although loss of hy-
droxyapatite content within the bone is observed in
BMES of the hip . A reasonable diagnosis of
Fig. 2. Secondary osteonecrosis. (A) T1-weighted and (B) fat suppressed T2-weighted images with two typical osteonecrosis
lesions, reactive interface, and concomitant BME in the patella and femoral condyle.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 323
BMES can be made by MRI only. Extensive, diffuse
BME involving an entire quadrant of the knee joint is
typical (see Fig. 1). An important differentiation from
other diseases with BME is the diffuse extension, the
lack of any other morphologic alterations, no history
of trauma, and the reversible course in most cases
. Spontaneous healing lasts from 3 to 12 months
(average, 6 months). In contrast to BMES of the hip,
a concomitant extensive joint effusion is uncommon
in BMES of the knee. A special migratory form can
be observed in rare cases [22,26]. After BME dis-
appears from an initially affected quadrant, new BME
may be observed in a different site in the knee (Fig. 4)
OCD usually affects the knee. OCD is the mani-
festation of osteonecrosi s in j uveniles, when the
growth plate is partly still open. The cause is probably
multifactorial and the consequence of abnormal ossi-
fication or focal stress combined with ischemia .
The prognosis and the course of the disease are much
better than for osteonecrosis in adults . OCD can
even be observed in the convex epiphysis of joints
and can be separated into five different stages. Unlike
secondary osteonecrosis in adults, the affected area in
OCD appears relatively small in most cases .
Initially on MRI a nonspecific subchondral BME
Fig. 4. Fat-suppressed T2-weighted coronal images showing migratory BMES. (A) Primary lesion (lateral). (B) Three months
later the lateral lesion has healed; there is new BME in the medial condyle.
Fig. 3. T2-weighted fat-suppressed images of stage II spontaneous osteonecrosis of the knee (SONK). (A) Coronal view.
(B) Sagittal view. Note the diffuse extension of the BME and the low signal line indicating subchondral fracture.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333324
(stage I) can be seen. Location, extension, and age of
the patient allow diagnosis in most cases. In stage II,
plain radiographs are still negative, but in MRI the
demarcation of a necrotic area from surrounding bone
is already visible (Fig. 5). In contrast to osteonecrosis
in older patients, stage II is still reversible. Stages III
through V can be diagnosed on plain radiographs.
Concomitant BME can occur in all stages, but it is not
as common as in osteonecrosis .
Chronic regional pain syndrome
CRPS is also known as algodystrophy, reflex
sympathetic dystrophy, or Morbus-Sudeck syndrome
. Following a trauma or i njury of unknown
origi n, a continuous burning pain, trophic distur-
bances, sensorimotor alterations, and, frequently,
psychic imbalance are observed. Initially one joint
is affected, but in chronic cases the entire extremity
may be involved. In CRPS three different stages
(acute, dystrophy, and atrop hy) can be sep arated
. In most cases, the diagnosis can be suspected
by history and clinical findings. In the initial stage,
bone scintigraphy shows intense tracer uptake by the
involved joint and periarticular tissues . On plain
radiographs, the earliest signs are typically patchy
structural changes after 6 to 12 weeks. MRI is not
the modality of choice in the diagnostic process of
CRPS , but it can contribute to diagnosis in ques-
tionable cases in the acute stage . Characteristic
findings on MRI in acute CRPS are diffuse BME on
both sides of the affected j oint and edematous
changes in periarticular soft tissue. In most cases,
joint effusion is visible [7,34]. In the initial stage, an
acute infection must be excluded. Differentiation
from edema caused by other disease is possible in
most cases. In migratory BMES, however, soft tissue
involvement is possible too. A continuous transition
from migratory BMES to CRPS may be possible .
Mechanical bone marrow edema
In almost all cases plain radiographs and CT allow
sufficient diagnosis in the acute phase following
injury to the extremity. Patients with nonspecific pain
that does not respond to therapy following a trauma
or overloading and who have negative plain radio-
graphs present a difficult diagnostic situation. Al-
though its specificity is very low, the advantage of
bone scintigraphy in occult trauma or mechanical
overload lies in its high sensitivity. In a consecutive
series of 176 patients, BME was detected on MRI in
72% after trauma . MRI significantly facilitates
diagnosis of posttraumatic or overloading pain.
Bone contusion (bone bruise)
Bone bruise is caused by direct injury to the bone.
On histologic evaluation, diffuse BME, microfrac-
tures of trabeculae, and hemorrhage can be found
. Bone contusions of the knee joint are a common
finding and are frequently observed after direct con-
tact, compression, or distraction injuries .By
definition, a bone bruise is not visible as a fracture
on plain-film radiographs or CT. In bone scintigra-
phy, increased tracer uptake in the affected area may
help diagnosis. Today, MRI is the modality of choice
for detecting bone contusions. A diffuse subcortical
BME is visible in the painful area with enhancement
Fig. 5. Stage II osteochondritis dissecans (OCD) with concomitant BME sagittal images. (A) T1-weighted image. (B) Fat-
suppressed T2-weighted image.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 325
after administration of contrast agents. Besides the
diffuse BME pattern, no signs of osteonecrosis or
fractures are visible on MRI (Fig. 6). A focal demin-
eralization zone may be observed on plain radio-
graphs 6 to 12 weeks after injury .
Microfractures are traumatic injuries of the bone
marrow in which cortical involvement is common.
There is no clear delimitation between bone bruise
and microfracture. Even in bone contusion of joint
structures, osteochondral microfractures may be pos-
sible. Plain-film radiographs are not helpful for
detecting microfractures. The fra cture line can be
detected on high-resolution CT, however. On T1-
weighted MRI, the microfracture is characterized by
a broad band of low signal in the bone marrow
coursing until the cortic al surface . Even on T2-
weighted MRI a thin band of low signal indicates
the fracture line surrounded by BME (Fig. 7) .
Sometimes the concomitant edema partially hides the
fracture line, and diagnosis is more difficult. In
compression fractures, extensive BME can usually
be observed. In contrast, distraction injuries may
show only minimal BME, and therefore misdiagnosis
with MRI is possible [39,40].
Stress-related bone marrow edema
Frequently, with mechanical stress or frontal mal-
alignment, subchondral BME of the overloaded com-
Fig. 6. Fat-suppressed T2-weighted images of bone bruise after anterior cruciate ligament rupture.(A) Sagittal view shows BME
of the lateral posterior tibia and central femoral condyle. (B) Coronal view of the same lesion.
Fig. 7. Microfracture: osteochondral microfracture not visible on radiographs. (A) Sagittal T1-weighted image. (B) Coronal fat-
suppressed T2-weighted image shows fracture in the lateral tibia plateau.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333326
partment can be observed . In most patients this
chronic mechanic overloading leads to progressive
early arthritic changes (subchondral sclerosis and
thinning of the hyaline cartilage) on radiographs.
Stress-related BME can be observed without any
arthritis in the knee, however. In MR examinations
in a healthy population without arthritis, subchondral
BME could be demonstrated after mechanical stress
combined with artificial malalignment of the mechan-
ical axis . On MRI, stress-related BME is char-
acterized by wedge-shaped subchondral BME in the
femur and tibia of the involved compartment with the
base of the wedge located at the site of the greatest
load (Fig. 8). Frequently, there are additional arthritic
signs with typical chondral and subchondral signal
changes, as discussed later. Therefore a continuous
transition from stress-related BME to activated oste-
oarthritis may be considered.
Stress fractures can be divided into fatigue and
insufficiency fractures. A fatigue fracture is caused
by repeated overloading of normal bony structures.
In contrast, insufficiency fractures occur sponta-
neously, without any trauma or overloading in patho-
logic, altered bony tissues (eg, osteoporotic bones)
. Differentiation between microfractures and
stress-related fractures is not possible with conven-
tional imaging modalities. The patient’s history can
help differentiate between microfracture (with a his-
tory of trauma) and stress fracture (with a history
Reactive bone marrow edema
Reactive BME occurs in a group of disorders in
which the underlying disease or a prior s urgical
procedure dominates the history, clinical findings,
prognosis, and course of the disease. BME in
these patients represents only a severe concomitant
component without any main influence on the ther-
apeutic management. In most cases there are no
essential differential diagnostic difficulties in separat-
ing reactive BME from the other types of BME de-
The most important diseases in which reactive
BME is noted are chronic polyarthritis, reactive
arthritis, bacterial arthritis, and osteomyelitis. Differ-
ential diagnosis is of utmost importance for ther-
apeutic management. In contrast to other imaging
modalities, MRI allows early detection, exact assess-
ment of bony involvement (location and extent),
evaluation of the severity of disease, and its differ-
entiation from other diseases . MRI should not be
used as the primary modality but should be used
when the diagnosis is questionable. In the initial stage
of chronic polyarthritis, MR imaging allows evalua-
tion of a joint effusion, synovial involvement, bony
erosions, periarticular soft tissue involvement, and
relatively early alterations of hyaline cartilage .
Intravenous administration of contrast agent may be
helpful for assessment of inflammatory activity .
In the acute phase of chronic polyarthritis a more or
less severe concomitant BME may be observed .
Fig. 8. Sagittal views of stress-related BME: lateral compartment osteoarthritis with valgus deformity and beginning
osteoarthritis. (A) T1-weighted image. (B) Fat-suppressed T2-weighted image.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 327
In most cases, however, a correct assignment of this
reactive BME is not difficult . Reactive arthritis
can be observed as a concomitant finding in several
diseases. It is important to differentiate reactive
arthritis from bacterial arthritis. Normally in reactive
arthritis there are no pathologic changes within the
bone marrow, but in rare cases BME can be observed
on MRI . Frequently, in these cases, differentia-
tion from an initial stage of chronic polyarthritis is
not possible by MRI .
In bacterial gonarthritis, MRI is helpful for evalu-
ating the involvement of cartilage, joint capsule, soft
tissue structures, and the bone marrow. MRI, however,
is indicated in unclear cases only . A concomitant
BME in bacterial arthritis must be conside red as
possible direct involvement of the bony structures
. In the initial stage of osteomyelitis, only diffuse
BME is visible, and differentiation between acute
infectious alterations within the bone and concomi-
tant BME may be difficult. In this situation intrave-
nous administration of contrast agent may facilitate
diagnosis . Changes in signal intensity are non-
specific, however, and it can be difficult to differen-
tiate concomitant BME from other forms of diffuse
BME. In most cases an exact diagnosis can be made
by history, clinical and laboratory findings, plain-film
radiographs, and appearance in MR imaging .
Assessment of osteomyelitis following to trauma or
surgery remains still a problem, however.
Normally MRI is not used f or diagnosis and
therapy planning in chronic degenerative joint dis-
eases. In unclear cases, however, MRI may be helpful
in detecting additional changes such as joint effusion,
subchondral edema, geodes, and reactive synovitis
(Fig. 9). Histologic evaluation of subchondral BME
in arthritic knee joints shows several pathologic
changes in the bone marrow . In a recent study
it has been shown that subchondral BME is correlated
to pain in patients with painful osteoarthritis of the
knee . A longitudinal follow-up study of these
patients over 30 months documented for the first
time that, besides mechanical malalignment, BME
represented the main risk factor for osteoarthritic
progression . There is no strong delimitation
between stress-related BME, described previously,
and subchondral BME in osteoarthritis.
Postoperative bone marrow edema
MRI is indicated for follow-up examinations after
surgery and when there is continued or recurrent pain
after surgery. In the knee joint postoperative BME is
frequently observed after reconstructions of liga-
ments, dri lling, and surgical procedures involving
the osteochondral compartment (Fig. 10). This reac-
tive BME can be seen up to 6 to 12 months after
surgery , and an appropriate diagnosis and as-
signment are not difficult in most cases. In patients
with persisting or recurrent pain after arthroscopies
with partial meniscectomies [52 –54] or ligament
reconstructions , subchondral signal alterations
have been described on MRI. In 94 patients with
meniscal tears who did not show any subchondral
signal alterations on MRI before surgery, a partial
meniscectomy was performed arthroscopically. MRI
Fig. 9. Coronal views of medial osteoarthritis. (A) T1-weighted image. (B) Fat-suppressed T2-weighted image shows BME in the
tibial plateau and focal in the femoral condyle.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333328
control examinations performed in all patients within
2 years revealed that 34% had femoral or tibial
subchondral BME in the region of the meniscectomy.
There was no correlation between the incidence of
BME and age, gender, or constitution of the hyaline
cartilage. The percentage of the removed meniscus
was identified as a risk factor, however . This
subchondral BME will disappear in many cases if
partial weight bearing is induced by early diagnosis
. The risk of developing manifest osteonecrosis
after partial meniscectomy seems to increase in
patients older than 50 years [53,56]. Clinical symp-
toms, course, and imaging findings of postoperative
BME after arthroscopy are the same as for SONK
. Furthermore, histologic examinations demon-
strated subchondral microfractures comp arable to
SONK in these cases . Predisposing factors may
be ischemia and local traumatizing during arthros-
copy as well as mechanic overloading (missing effect
of shock absorption and further deterioration of the
mechanical axis because of the removal of the me-
Tumors and tumorlike lesions of the knee joint
Benign and malignant tumors of the knee joint are
fairly common. After initial plain radiographs, MRI
with contrast administration should be performed for
exact assessment of bone involvement and involve-
ment of soft tissue structures . MRI morphology
of tumors in the knee joint is the same as in other
joints. Concomitant reactive BME can be observed in
almost all stages of various tumors . In most
cases better delineation of concomitant edema from
tumor tissue is possible by using intravenous contrast
agent . Characteristic MRI findings of tumors
together with specific plain-film abnormalities allow
a clear differentiation from other entities with BME in
almost all cases .
Therapeutic management of BME depends es-
sentially on the disease that causes the BME. Non-
steroidal anti-inflammatory drug (NSAID) or pain
medications have shown only a limited effect, espe-
cially for the night pain. Mechanical unloading by
partial weight bearing or drilling the edematous bone
may lead to pain relief . An interesting new
treatment is the medication therapy with Iloprost
(Ilomedin, Schering, Berlin, Germany), a prostacyclin
analogue . The effect of iloprost has been evalu-
ated in two prospective MRI studies in patients with
painful BME of the knee (BMES, bone bruise, stress-
related BME, and reactive BME with osteoarthritis).
The preliminary data have shown clinical success for
pain relief and rapi d reg ression of the BME and
subchondral lesions on MRI . Iloprost may be a
successful therapeutic approach for patients with
painful BME in the future.
Ischemic bone marrow edema
In osteonecrosis differentiating secondary osteo-
necrosis from SONK is important for therapeutic
decision making. In secondary osteonecrosis, de-
creasing intraosseous pressure by core decompres-
sion, a minimal surgical procedure, can lead to
immediate pain relief . Repair of manifest necro-
sis (irreversible stage II osteonecrosis) is not possible
by drilling only, however . In early-stage osteo-
necrosis without joint space destruction (stages I–II),
core decompression has shown clinical success in
79% with an average follow-up of 7 years .
Fig. 10. Postoperative BME: sagittal views 6 weeks after osteochondral transplantation (mosaicplasty). (A) T1-weighted image.
(B) Fat-suppressed T2-weighted image shows perifocal BME around the cylindrical transplants.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 329
Because osteonecrotic areas are relatively large, and
multiple sites are involved, a bone–cartilage trans-
plantation is not a reasonable alt ernative in most
cases. After an osteochondral fracture and clinical
and radiographic progression (stages III and IV), total
knee arthroplasty is recommended . In SONK
with subchondral BME in the initial stage, a prog-
nostic assessment is important. Lesions that involve
more than 50% of the bone in the joint have a
worse prognosis . In a retrospective MRI study
of 23 patients who received no treatment, the pres-
ence of a subchondral hypointense line more than
4 mm thick or 14 mm long on T2-weighted images
indicated a significant risk for early osteochondral
fracture and progression (see Fig. 3) .Inthe
initial stage, conservative therapy with partial weight
bearing can be recommended, because in many cases
the necrosis may stabilize. In pat ients with risk
factors or flattening of the fem oral condyle an d
mechanical malalignment, a realignment osteotomy
should be consi dered, depending on the patient’s
age . Additionally de
bridement, curettage, and
drilling can be tried arthroscopically . In most
cases cartilage transplantation seems not to be rea-
sonable because of the patient’s age. A good thera-
peutic solution in advanced stages is a minimally
invasive unicondylar prosthesis .
In BMES the aim of treatment is pain relief and
shortening of the spontaneous course. NSAID and
pain medications are not very successful. Mechanic
unloading by partial weight bearing decreases com-
plaints but has only a limited effect on night pain.
In BMES of the hip joint, core decompression has
demonstrated immediate pain relief and significant
shortening of the spontaneous clinical course .
For the knee joint similar data are not available. In
OCD the therapeutic considerations are not influ-
enced by concomitant BME. In stage I and II OCD
conservative therapy (mechanical unloading by par-
tial weight bearing) or drilling in selected cases seems
to be reasonable. In lat e stages surgical t herapy
includes simple drilling, stabilization of the fragment,
curettage combined with drilling, and lastly osteo-
chondral or chondrocyte transplantation . In the
therapeutic management of CRPS several considera-
tions must be taken in account. Sympathiolytic
agents, various drugs, and physiotherapy have been
recommended with varying degrees of success .
Mechanical bone marrow edema
In bone bruises the primary goal of treatment
is symptomatic pain relief by mechanical unloading
and interruption of sports activities for at least
6 weeks. After pain is relieved and before full weight
bearing resumes, a control examination with MRI
can be recommended . In stress-related BME
the main goal of therapy is relief of pain by NSAID
or analgesics as well as mechanical unloading by
partial weight bearing. Pat ients with stress-related
BME combined with malalignment of the mechani-
cal axis show a great risk for decompensation of
the compartment, and therefore realignment should
be considered in patients younger than 60 years of
age. The treatment of microfractures and stress frac-
tures is similar to the treatment for bone bruises, but
unloading is recommended for 6 to 12 weeks. Fur-
thermore, the cause of a stress fracture should be
evaluated, and prophylactic recommendations should
Reactive bone marrow edema
Reactive BME represents only a more or less
severe concomitant component without any main
influence to the therapeutic management. Only in
osteoarthritis of the knee does subchondral BME
represent an important risk factor for progression,
indicating the beginning decompensation of the joint
. Therefore arthroscopic de
bridement in osteo-
arthritis should be combined with decompression
and/or mechanical realign ment of the mechanical
axis. In patients with osteoarthriti s who are older
than 60 years, the indication for arthroscopy should
be restrictive, because there is a high risk of less
postoperative pain relief, rapid progression of osteo-
arthritis, and eventually development of postopera-
tive SONK [18,52,54].
BME is a common finding when patients with
knee pain are evaluated by MRI. The typical MRI
signal patterns for BME are nonspecific, however,
and occur in several diseases of the knee. This article
categorizes painful BME of the knee joint into three
distinct etiological groups: ischemic BME (osteone-
crosis, OCD, BMRS, and CRPS), mechanical BME
(bone bruise, microfracture, stress-related BME, and
stress fracture), and reactive BME (inflammatory
gonarthritis, degenerative gonarthrosis, postoperative
BME, and tumor-related BME). The therapeu tic
concepts have been described briefly in a short
overview of the different therapeutic approaches.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333330
 Vahlensieck M, Reiser M. MRT des Bewegungsappa-
rates. Stuttgart: Thieme Verlag; 1997.
 Munshi M, Davidson M, MacDonald PB, Froese W,
Sutherland K. The efficacy of magnetic resonance im-
aging in acute knee injuries. Clin J Sport Med 2000;
 H uber W, Trieb K. Serious consequences of the
wrong diagnosis of meniscal lesion in a case of stress
fracture of the distal femur. Arthroscopy 2002;18(8):
935 – 8.
 Hofmann S, Kramer J, Breitenseher M, Aigner N. The
painful bone marrow edema of the knee [Das schmerz-
hafte Knochenmarko¨dem im Kniegelenk]. Arthrosko-
pie 2003;16(2):88 –100.
 Beltran J, Shankman S. Magnetic resonance imaging
of bone marrow disorders of the knee. Magn Reson
Imaging Clin N Am 1994;2(3):463 – 73.
 Fertakos RJ, Swayne LC, Colston WC. Three-phase
bone imaging in bone marrow edema of the knee. Clin
Nucl Med 1995;20(7):587 – 90.
 Hofmann S. The bone marrow oedema pattern in the
syndromes of transient osteopor osis, algodystrophy
and osteonecrosis. J Bone Joint Surg Suppl [Br] 1998;
 Plenk Jr H, Hofmann S, Breitenseher M, Urban M.
[Pathomorphological aspe cts and repair mechanisms
of femur hea d necrosis]. Orthopade 20 00;29(5):
 Plenk Jr H, Hofmann S, Eschberger J, Gstettner M,
Kramer J, Schneider W, et al. Histomorphology and
bone morphometry of the bone marrow edema syn-
drome of the hip. Clin Orthop 1997;334:73– 84.
 Felson D, Chaisson CE, Hill CL, Totterman S, Gale D,
Skinner HB, et al. The asso ciation of bone marrow
lesions with p ain in knee osteoarthritis. Ann Intern
Med 2001;134:541– 9.
 Arnoldi CC, Lemperg R. Intraosseous hypertension
and pain in the knee. J Bone Joint Surg [Br] 1975;57:
360 – 5.
 Hofmann S, Schneider W, Breitenseher M, Urban M,
Plenk Jr H. Die ‘‘transiente osteoporose’’ als reversible
Sonderform der Hu
ftkopfnekrose. Orthopade 2000;
29(5):411 – 9.
 Major NM, Helms CA. MR imaging of the knee: find-
ings in asym ptomatic collegiate basketball players.
AJR Am J Roentgenol 2002;179(3):641 – 4.
 Patel DV, Breazeale NM, Behr CT, Warren RF, Wick-
iewicz TL, O’Brien SJ. Osteonecrosis of the knee: cur-
rent clinical conce pts. Knee Surg Sports Traumatol
Arthrosc 1998;6(1):2 –11.
 Ecker ML, Lotke PA. Osteonecrosis of the medial part
of the tibial plateau. J Bone Joint Surg [Am] 1995;77(4):
 Khanuja H, Rajadhyaksha AD, Etienne G, Hungerford
MW, Jones LC, H ungerford DS, et al. Av ascular
necrosis of the knee: a comprehensive review. Pro-
ceedings of the annual meeting of the AAOS. Rose-
mont: Amer ican Academy of Orthopedic Surgeons;
2003. p. 404.
 Mont MA, Baumgarten KM, Rifai A, Bluemke DA,
Jones LC, Hungerford DS. Atraumatic osteonecrosis
of the knee. J Bone Joint Surg [Am] 2000;82(9):
 Pape D, Seil R, Fritsch E, Rupp S, Kohn D. Prevalence
of spontaneous osteonecrosis of the medial femoral
condyle in elderly patients. Knee Surg Sports Trauma-
tol Arthrosc 2002;10(4):233 – 40.
 Sokoloff RM, Farooki S, R esnick D. Spontaneous
osteonecrosis of the knee associated with ipsilateral
tibial plateau stress fracture: report of two patients
and review o f t he literature. Skeletal Radiol 2001;
 Yamamoto T, Bullough PG. Spontaneous osteonecrosis
of the knee: the result of subchondral insufficiency
fracture. J Bone Joint Surg [Am] 2000;82(6):858 – 66.
 Wilson AD, Murphy WA, Hardy DC, Totty WG. Tran-
sient osteoporosis: transient bone marrow edema. Ra-
 Lakhanpal S, Ginsburg WW, Luthra HS, Hunder GG.
Transient regional osteoporosis. A study of 56 cases
and r eview of the literature. Ann Intern Med 1987;
 Hofmann S, Engel A, Neuhold A, Leder K, Kramer J,
Plenk Jr H. Bone-marrow oedema syndrome and tran-
sient osteoporosis of the hip. An MRI-controlled study
of treatment by core decompression. J Bone Joint Surg
[Br] 1993;75(2):210 – 6.
 Koo KH, Dussault R, Kaplan PA, Kim R, Ahn JI,
Christopher J, et al. Age-rela ted marrow conversion
in the proximal metaphys is of the femur: evaluation
with T1-weighted MR imaging. Ra diology 1998;
 Turner DA, Templeton AC, Selzer PM, Rosenberg AG,
Petasnick JP. Femoral capital osteonecrosis: MR find-
ing of diffuse mar row abnormalities without focal
lesions. Radiology 1989;171(1):135 –40.
 Glockner JF, Sundaram M, Pierron RL. Radiologic
case study. Trans ient migratory osteoporosis of the
hip and knee. Orthopedics 1998;21(5):594 – 600.
 Wambeek N, Munk PL, Lee MJ, Meek RN. Intra-ar-
ticular regional migratory osteoporo sis of the knee.
Skeletal Radiol 2000;29(2):97 – 100.
 Gaeta M, Mazziotti S, Minutoli F, Vinci S, Blandino A.
Migrating transient bone marrow edema syndrome of
the knee: MRI findings in a new case. Eur Radiol 2002;
12(Suppl 4):S40 – 2.
 Kramer J, Hofmann S. MRT der Osteochondro-
sis dissecans des Kniegelenks. In: Imhoff A, editor.
Fortbildung orthopa¨ die 5: magnetresonanztomo-
graphie. Darmstad t (Germany): Steinkopf Verlag;
2001. p. 70 – 5.
 Hofmann S, Kramer J, Leder K, Plenk Jr H, Engel A.
[The non-traumatic femur head necrosis in the adult. I:
pathophysiology, clinical picture and therapeutic
options]. Radiologe 1994;34(1):1 –10.
 Roemer FW, Bohndorf K. Long-term osseous sequelae
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 331
after acute trauma of the knee joint evaluated by MRI.
Skeletal Radiol 2002;31(11):615 – 23.
 Hogan C, Hurwitz S. Treatment of complex regional
pain syndrome of the lower extremity. J Am Acad
Orthop Surg 2002;10:281– 9.
 Fialka V. The diagnosis of reflex sympathetic dystro-
phy. European Journal of Physical Medicine and Re-
 Schimmerl S, Schurawitzky H, Kramer J, Imhof H,
Fialka V. Imaging modalities for diagnosis of reflex
sympathetic dystrophy. European Journal of Physical
Medicine and Rehabilitation 1992;2:2 – 7.
 Rangger C, Kathrein A, Frey C, Klestil T, Kreczy A.
Bone bruise of the knee: histology and cryosections in
5 cases. Acta Orthop Scand 1998;69:291 – 4.
 Hayes CW, Brigido MK, Jamadar DA, Propeck T.
Mechanism-based pattern approach to classification
of complex injuries of the knee depicted at MR
imaging. Radiographics 2000;20(Special issue):
 Sanders TG, Medynski MA, Feller JF, Lawhorn KW.
Bone contusion patterns of the knee at MR imaging:
footprint of the mechanism of injury. Radiographics
2000;20(Special issue):S135– 51.
 Breitenseher MJ, Gaebler C. Trauma of the wrist. Eur J
Radiol 1997;25(2):129– 39.
 Palmer WE, Levine SM, Dupuy DE. Knee and shoul-
der fractures: association of fracture detection and mar-
row edema on MR images with mechanism of injury.
Radiology 1997;204(2):395– 401.
 Weber WN, Neumann CH, Barakos JA, Petersen SA,
Steinbach LS, Genant HK. Lateral tibial rim (segond)
fractures: MR imaging characteristics. Radiology
 Felson D, McLaughlin S, Goggins J, LaValley MP,
Gale ME, Totterman S, et al. Bone marrow edema
and its relation to progression of knee osteoarthritis.
Ann Intern Med 2003;139:330 – 6.
 Schweitzer ME, White LM. Does altered biome-
chanics cause marrow edema? Radiology 1996;
 Daffner RH, Pavlov H. Stress fractures: current con-
cepts. AJR Am J Roentgenol 1992;159:245 – 52.
 Fritz K, Raith J, Hermann J. MRT bei der Polyarthritis.
In: Imhoff A, editor. Fortbildung orthopa¨die 5: magnet-
resonanztom ographie. Darmstadt: Steinkopff Verlag;
2001. p. 35– 44.
 Zacher J, Reiser M. Ergebnisse der Kernspintomogra-
phie bei chronisch-entzu
gen – Korrelation mit intraoperativen Befunden. Akt
Rheumatol 1985;10:195– 201.
 Poleksic L, Musikic P, Zdravkovic D, Watt I, Bacic G.
MRI evaluation of the knee in rheumatoid arthritis. Br
J Rheumatol 1996;35(Suppl 30):36 – 9.
 Yu JS, Chung C, Recht M, Dailiana T, Jurdi R. MR
imaging of tophaceous gout. AJR Am J Roentgenol
 Lee S, Suh K, Kim YW. Septic arthritis versus tran-
sient synovitis at MRI: preliminary assessment with
signal intensity alterations in bone marrow. Radiology
1999;211:459 – 65.
 Tang J, Gold R, Bassett L, Seeger L. Musculoskeletal
infections of the extremities: Evaluation with MR im-
aging. Radiology 1988;166:205–9.
 Link TM, Steinbach LS, Ghosh S, Ries M, Lu Y, Lane
N, et al. Osteoarthritis: MR imaging findings in differ-
ent stages of disease and correlation with clinical find-
ings. Radiology 2003;226(2):373 –81.
 Herber S, Runkel M, Pitton MB, Kalden P, Thelen M,
Kreitner KF. [Indirect MR-arthrography in the follow
up of autologous osteochondral transplantation]. Rofo
Fortschr Geb Rontgenstr Neuen Bildgeb Verfahr 2003;
 Kobayashi Y, Kimura M, Higuchi H, Terauchi M,
Shirakura K, Takagishi K. Juxta-articular bone marrow
signal changes on magnetic resonance imaging follow-
ing arthroscopic meniscectomy. Arthroscopy 2002;
 Muscolo DL, Costa Paz M, Makino A, Ayerza MA.
Osteonecrosis of the knee following arthroscopic
meniscectomy in patients over 50 years old. Arthros-
copy 1996;12(3):273 – 9.
 Santori N, Condello V, Adriani E, Mariani PP. Osteo-
necrosis after arthroscopic medial meniscectomy. Ar-
throscopy 1995;11(2):220 – 4.
 Athanasian EA, Wickiewicz TL, Warren RF. Osteone-
crosis of the femoral condyle after arthroscopic recon-
struction of a cruciate ligament. Report of two cases.
J Bone Joint Surg [Am] 1995;77(9):1418 – 22.
 Prues-Latour V, Bonvin JC, Fritschy D. Nine cases of
osteonecrosis in elderly patients following arthroscopic
meniscectomy. Knee Surg Sports Traumatol Arthrosc
 Faletti C, Robba T, de Pe tro P. Post menis cect omy
osteonecrosis. Arthroscopy 2002;18(1):91– 4.
 Nakamura N, Horibe S, Nakamura S, Mitsuoka T. Sub-
chondral microfr actur e o f the knee without osteo-
necrosis after arthroscopic medial meniscectomy.
Arthroscopy 2002;18(5):538– 41.
 Ishida T, Goto T, Motoi N, Mukai K. Intracortical
chondroblastoma mimicking intra-articular osteoid os-
teoma. Skeletal Radiol 2002;31(10):603 – 7.
 Aigner N, Steinbo¨ck G, Schneider W, Krasny C, Land-
siedl F. Treatment of bone marrow edema of the talus
with the prostacyclin analogue iloprost: an MRI-con-
trolled investigation of a new method. J Bone Joint
Surg [Br] 2001;83:855 –8.
 Aigner N, Vakil-Adli A, Hofmann S, Kramer J, Mayer-
ho¨fer M, Meizer R, et al. Double-blind, randomized
controlled 4-week study of iloprost in pa tients with
painful bone marrow edema of the knee [ abstract].
Association Research Circulation Ossseous (ARCO)
Symposium, October 9 – 11, 2003, South Korea.
 Lecouvet FE, van de Berg BC, Maldague BE, Lebon
CJ, Jamart J, Saleh M, et al. Early irreversible osteo-
necrosis versus transient lesions of the femoral con-
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333332
dyles: prognostic value of subchondral bone and mar-
row changes on MR imaging. AJR Am J Roentgenol
 Marti CB, Rodriguez M, Zanetti M, Romero J. Spon-
taneous osteonecrosis of the medial compartment of the
knee: a MRI follow-up after conservative and operative
treatment, preliminary results. Knee Surg Sports Trau-
matol Arthrosc 2000;8(2):83– 8.
 Schneider T, Fink B, Jerosch J, Assheuer J, Ru
The value of MRI as postoperative control after ar-
throscopic treatment of osteochondritis dissecans. Arch
Orthop Trauma Surg 1998;117:235– 9.
S. Hofmann et al / Orthop Clin N Am 35 (2004) 321–333 333