Clinical Radiology 71 (2016) 570e575

Contents lists available at ScienceDirect

Clinical Radiology
journal homepage: www.clinicalradiologyonline.net

Cervical facet oedema: prevalence, correlation to

symptoms, and follow-up imaging
M.T. Nevalainen a, b, *, P.J. Foran a, J.B. Roedl a, A.C. Zoga a, W.B. Morrison a
a
Division of Musculoskeletal Imaging and Intervention, Department of Radiology, Thomas Jefferson University
Hospital, Sidney Kimmel Medical College at Thomas Jefferson University, 132 South 10th Street, Philadelphia,
PA 19107, USA
b
Department of Diagnostic Radiology, Oulu University Hospitaļ P.O. Box 50, 90029, Oulu, Finland

article in formation AIM: To evaluate the prevalence of cervical facet oedema in patients referred for magnetic
resonance imaging (MRI) to investigate neck pain and/or radiculopathy, and to investigate
Article history: whether there is a correlation between the presence of oedema and patients’ symptoms.
Received 27 December 2015 MATERIALS AND METHODS: A retrospective report review of 1885 patients undergoing
Received in revised form cervical spine MRI between July 2008 and June 2015 was performed. Exclusion criteria
22 February 2016 included acute trauma, surgery, neoplastic disease, or infection in the cervical spine. One
Accepted 29 February 2016 hundred and seventy-three MRI studies with cervical facet oedema were evaluated by each of
the two radiologists. In these patients, the grade of bone marrow oedema (BMO) and corre-
sponding neuroforaminal narrowing at the cervical facets was assessed. Correlation with
symptoms was performed based on pre-MRI questionnaire.
RESULTS: The prevalence of cervical facet oedema was 9%; the most commonly affected
levels were C3e4, C4e5, and C2e3. A total of 202 cervical facets were evaluated: mild BMO
was seen in 35%, moderate in 41%, and severe in 24% of cases. Surrounding soft-tissue oedema
was observed in 36%, 69%, and 92% of the BMO grades, respectively. The correlations between
unilateral radiculopathy and ipsilateral facet BMO grades were 79%, 83%, and 73% (chi-square,
p<0.001), respectively. Furthermore, neuroforaminal narrowing on the corresponding level
was found in 35%, 38%, and 11% of cases, respectively. At follow-up imaging, facet oedema was
most likely to remain unchanged or to decrease.
CONCLUSION: The prevalence of cervical facet oedema is 9%. Cervical facet oedema is
associated with ipsilateral radiculopathy. Neuroforaminal narrowing, however, is not associ-
ated with facet oedema.
Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Introduction

Neck pain is a very common finding causing substantial

*Guarantor and correspondent: M.T. Nevalainen, Division of Musculo-
skeletal Imaging and Intervention, Department of Radiology, Thomas
Jefferson University Hospital, Sidney Kimmel Medical College at Thomas
Jefferson University, 132 South 10th Street, Philadelphia, PA 19107, USA.
Tel.: þ1 215 955 0420; fax: þ1 215 923 1562.
E-mail address: mikaneva@paju.oulu.fi (M.T. Nevalainen).
disability in the general population, with on average half of
all people suffering from it within their lifetime.1,2 Many
reasons behind neck pain and upper extremity radiculop-
athy can be distinguished including mechanical impinge-
ment of the spinal cord and nerve roots by spondylolisthesis,

http://dx.doi.org/10.1016/j.crad.2016.02.026
0009-9260/Ó 2016 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.
 M.T. Nevalainen et al. / Clinical Radiology 71 (2016) 570e575
disc herniation, posterior vertebral body osteophytic spur-
ring, and osseous hypertrophy related to uncovertebral joint
and facet joint arthropathy.3 Cervical facet joints have been
also shown to be a source of pain in the neck, head, and
upper extremity4e6 mainly through facet joint osteoar-
thritis, which based on cadaveric studies, is most common in
the upper-mid cervical region at C2e3, C3e4, and C4e57. In
cervical radiculopathy, spondylosis resulting in neuro-
foraminal narrowing contributes to 70% of cases.8 Decreased
disc height and degenerative changes of the uncovertebral
joints anteriorly or cervical facet joints posteriorly are also
common contributors to radiculopathy.9 Usually symptoms
related to radiculopathy tend to be unilateral, and bilateral
symptoms are more consistent with arthritis of the cervical
spine.8 Concerning neuroforaminal narrowing, the magnetic
resonance imaging (MRI) findings should be ultimately
correlated with clinical findings, because both false-positive
and false-negative rates have been shown to be high.10
As a reference standard, MRI of the cervical spine is
commonly performed to evaluate the cause of neck pain
and upper extremity radicular symptoms. In the authors’
clinical practice, cervical facet oedema has been observed,
i.e., bone marrow oedema (BMO) adjacent to facet joints, on
MRI cervical spine studies performed to evaluate the cause
of neck pain and/or upper extremity radicular symptoms.
This finding has been included in the MRI reports, but its
clinical significance is uncertain. The purpose of the present
study was to determine the prevalence of cervical facet
oedema seen on cervical spine MRI. The severity of the BMO
and associated neuroforaminal narrowing at cervical facets
in MRI was graded and correlated to the severity of the BMO
with radicular symptoms. Additionally, follow-up MRI
studies were examined to evaluate the course of cervical
facet oedema.
Materials and methods
Patients
This study was carried out in accordance with the Code
of Ethics of the World Medical Association (Declaration of
Helsinki). Institutional review board approval was obtained
and the requirement for informed consent was waived. A
retrospective report review of 1885 patients undergoing
cervical spine MRI between July 2008 and June 2015 was
performed with a keyword search using the terms “facet”
and “oedema” from the picture archiving and communica-
tion system (PACS). These examinations were originally
read by fellowship-trained musculoskeletal radiologists.
Patients referred for MRI for evaluation of neck pain and/or
upper extremity radiculopathy with a patient history sheet
available were included. Exclusion criteria included acute
trauma, prior cervical spine surgery, neoplastic disease in
the cervical spine, and infection in the cervical spine or
paravertebral tissues. Positive studies were reviewed by a
fellowship-trained musculoskeletal radiologist to confirm
the presence of cervical facet oedema. Ultimately, this
yielded 173 patients (mean age 61 years, range 30e90
571
years) with cervical facet oedema. Moreover, follow-up MRI
was available for 25 patients. For these patients, the follow-
up time in months was collected and the grade of BMO at
the cervical facet was evaluated.
MRI and image analysis
All cervical spine MRI was performed using 1.5 T MRI
systems (Optima and Signa, General Electric Medical Sys-
tems, Milwaukee, WI, USA) with dedicated coils and routine
protocols in axial and sagittal planes with T1, T2, and short
tau inversion recovery (STIR) or T2 fat-saturated (FS) se-
quences. Cervical facet oedema was defined as hyperintense
(bright) signal on either STIR or T2 FS sequences in the
articular processes either side of the facet joints. The normal
bone marrow signal at the adjacent facet joints was used as
a reference. Grading of the BMO was performed as follows:
mild ¼ faint oedema, only part of the pars interarticularis
affected; moderate ¼ clearly recognisable oedema, part or
all of the pars interarticularis affected; severe ¼ intense
oedema, the whole of the pars interarticularis affected on
both sides of facet. The term BMO is used in this paper to
describe the radiological finding of hyperintense (bright)
signal on either STIR or T2 FS sequences, even though it has
been shown that histologically no oedema is present.11
As part of the study, 100 cervical spine MRI studies that
were not reported as having BMO were evaluated by both
readers to evaluate for false-negative studies. Neuro-
foraminal narrowing at the affected cervical level was
graded as none, mild (less than one-third narrowed),
moderate (less than two-thirds narrowed), or severe (more
than two-thirds narrowed). The patients’ age, gender, the
grade of neuroforaminal narrowing, and the grade of BMO
at the cervical facets between C2 to T1 on sagittal STIR or T2
FS images of the cervical spine were recorded. Each of two
fellowship-trained musculoskeletal radiologists, who were
blinded to the clinical data, reviewed the MRI images. In
cases of disagreement, a third musculoskeletal radiologist
made the final decision.
Clinical data
Of patients with cervical facet oedema, a pre-MRI ques-
tionnaire completed by the patient was assessed to collect
both illustrated and written information on neck pain and
upper extremity radiculopathy. The side of the neck pain
was recorded as midline, right, or left, and for radiculopathy
either right or left. In statistical analysis unilateral symp-
toms were defined either right- or left-sided neck pain and/
or radiculopathy.
Statistical analysis
The association between radiculopathy symptoms and
same-side cervical facet oedema was evaluated using the
chi-square test. The ManneWhitney test was applied to test
differences between soft-tissue oedema groups. The cut-off
for neuroforaminal narrowing was defined as at least
moderate or severe narrowing for statistical purposes. A
kappa value was applied to assess inter-reader variability.
572 M.T. Nevalainen et al. / Clinical Radiology 71 (2016) 570e575

Statistical software (SPSS, version 22.0, Chicago, IL, USA) Table 1

was used for the analysis. The distribution per cervical spine level of the 202 cervical facets with BMO
in 173 patients.

Results Cervical spine level Number of cervical facet joint Percent (%)
with adjacent BMO
C2e3 39 19.3
Demographics of cervical facet oedema
C3e4 70 34.7
C4e5 49 24.3
Among 1885 patients who underwent cervical spine C5e6 13 6.4
MRI, 173 had cervical facet BMO representing a prevalence C6e7 13 6.4
C7eT1 18 8.9
of 9%. The average age of patients with cervical facet
Total 202 100
oedema was 61 years, and 49% were male. BMO was seen at
a total of 202 cervical facets in the 173 patients with cervical BMO, bone marrow oedema.

facet oedema. Cervical facet oedema was observed adjacent

to one joint only in 148/173 (86%) patients, adjacent to two
facet joints in 21/173 (12%) patients, and adjacent to three
facet joints in 4/173 (2%) patients. Fig 1 shows examples of
cervical facet oedema in sagittal and axial MRI planes.
Bilateral facet oedema was seen in 21/173 (12%) patients. In
this subgroup, the BMO was seen bilaterally at the same
cervical level in nine out of 21 patients. In cases of unilateral
BMO, the most commonly affected level was C3e4, which
accounted for 70/202 (35%) cases, followed by C4e5 with
49/202 (24%) cases, and C2e3 with 39/202 (19%) cases.
Table 1 summarises the distribution of facet oedema on
cervical spine levels.
Grading of cervical facet oedema and correlation to
radiculopathy
Using the study group of 173 patients, further classifi-
cation of the BMO either into mild, moderate, or severe
BMO was performed. Accordingly, 202 cervical facets with
oedema were assessed: mild BMO was seen in 70/202 (35%)
cases, moderate in 84/202 (41%) cases, and severe in 48/202
(24%) cases. Inter-reader reliability was substantial
(kappa¼0.75) for grading the BMO. Adjacent soft-tissue
oedema was observed in patients with mild BMO in 25/70
(36%) cases, with moderate BMO in 58/84 (69%) cases, and
with severe BMO in 44/48 (92%) cases. The difference
between soft-tissue oedema groups was statistically sig-
nificant (p<0.001, ManneWhitney). Fig 2 demonstrates
examples of different grades of BMO at cervical facets.
When assessing patients with unilateral radiculopathy, a
strong correlation was found between the side of symptoms
and the side of cervical facet oedema. Of the 173 patients
with facet oedema, 147 (85%) had lateralising neck and/or
arm pain and 26 (15%) reported midline neck pain only. Of
the 147 patients with lateralising symptoms, 34 had bilat-
eral symptoms and 113 patients had unilateral symptoms.
After excluding patients with bilateral facet oedema, 100
patients with unilateral symptoms and unilateral facet
oedema remained of whom 79/100 (79%, p<0.001, chi-
square) had facet oedema ipsilateral to the symptoms.
The association between the severity of facet oedema
and the presence of ipsilateral symptoms was subsequently
examined. Accordingly, patients with only unilateral
symptoms in each subgroup (BMO graded mild, moderate,
or severe) were studied. There was a 79% (26/33) correla-
tion with mild BMO, an 83% (34/41) correlation with mod-
erate BMO, and a 73% (19/26) correlation with severe BMO
in the cervical facets (Table 2). A correlation between uni-
lateral symptoms and BMO was observed, and neuro-
foraminal narrowing was also seen on that level. The rate of
cases with corresponding neuroforaminal narrowing on the

Figure 1 (a) A 63-year-old man with cervical facet oedema at the left C2e3 facet joint (arrow) with adjacent soft-tissue oedema observed on the
sagittal T2 FS MRI sequence. (b) The axial image (T2 FS) shows BMO at the left C3e4 facet joint (arrow) in a 67-year-old male patient.
 M.T. Nevalainen et al. / Clinical Radiology 71 (2016) 570e575 573

Figure 2 Grading of cervical facet oedema. (a) Sagittal MRI demonstrates mild BMO at the left C2e3 facet joint (arrow) in a 48-year-old woman.
(b) Moderate BMO is observed in the left C2e3 facet joint (arrow) in a 63-year-old man. (c) Severe BMO is seen at the right C3e4 facet joint
(arrow) in a 47-year-old woman. All cases (aec) show adjacent soft-tissue oedema posterior to the BMO.

same level was 35% (9/26) in the mild BMO subgroup, 38% course of facet oedema. Unfortunately, the clinical data
(13/34) in moderate BMO subgroup, and 11% (2/19) in se- were insufficient or non-specific to make any correlation to
vere BMO subgroup (Table 2). When assessing neuro- the neck pain or radiculopathy with follow-up imaging.
foraminal narrowing in general with the BMO in the Fig 3 shows an example case with decreasing cervical facet
cervical facets, the rate of cases with foraminal narrowing oedema on follow-up MRI.
on the same level as the BMO was 19/70 (27%) with mild
BMO, 32/84 (38%) with moderate BMO, and 13/48 (27%) Discussion
with severe BMO. Thus, no correlation was found between
the severity of BMO and the presence of neuroforaminal A multitude of disease entities can cause axial neck pain
narrowing. and upper extremity radiculopathy, one of them being
cervical facet degeneration; this is common with increasing
Follow-up imaging of cervical facet oedema
prevalence with age, becoming nearly universal in patients
>60 years.12 Although frequently asymptomatic, cervical
Follow-up MRI was available for 25 patients with cervical
facet arthropathy has been demonstrated to be a source of
facet oedema. In total, these patients had 27 facets with
neck pain.4e6 The prevalence of facet oedema has not been
BMO. The average follow-up time was 22 months (range
studied extensively, but Friedrich et al. (2007)13 found BMO
1e58 months). In 10/27 cervical facets the intensity of BMO
and surrounding soft-tissue oedema at the lumbar facets in
stayed the same, in 12/27 facets the intensity of BMO
21 of 145 (14%) patients referred for investigation of lower
decreased, and in 5/27 facets the intensity of BMO
back pain. Correlation to any symptoms was not performed,
increased. The average follow-up times for each subgroup
so the significance of the BMO at the facets remained un-
were 16, 28, and 20 months, respectively. Moreover, the
known.13 Based on cadaveric studies, osteoarthritic changes
grade of the BMO at the facets did not play a role in the
at cervical facet joints seem to be most common at the
upper-mid cervical levels (C2e3, C3e4, and C4e5) levels,7
Table 2 but no imaging studies have verified these results.12 These
The association between the grade of BMO and corresponding ipsilateral findings are in concordance with the present authors’
radicular symptoms.
finding that cervical facet oedema is observed in the upper-
Association Number of p-Valuea Neuroforaminal mid cervical spine in 78% of cases (Table 1).
patients narrowing on In the present study, the prevalence of cervical facet
the same levelb
oedema was 9% in patients referred for MRI of the cervical
Unilateral symptoms to mild 26/33 (79%) <0.001 9/26 (35%) spine for investigation of neck pain and/or upper extremity
same-side BMO
Unilateral symptoms to 34/41 (83%) <0.001 13/34 (38%)
radiculopathy. A significant 79% correlation was observed
moderate same-side BMO between the side of facet oedema and the side of symptoms
Unilateral symptoms to 19/26 (73%) <0.001 2/19 (11%) in patients with unilateral lateralising symptoms and uni-
severe same-side BMO lateral cervical facet oedema. Possible reasons for this cor-
All correlations were statistically significant. Additionally, neuroforaminal relation include the involved joints being a direct source of
narrowing was not associated with the severity of BMO and reported pain (particularly in patients who reported unilateral neck
symptoms on that cervical level. pain), an inflammatory element to the facet joint arthrop-
BMO, bone marrow oedema.
a
Chi-square test.
athy causing irritation of the traversing nerve roots. It is also
b
Neuroforaminal narrowing was defined here as at least moderate or possible that the BMO is a surrogate marker for facet joint
severe narrowing. arthropathy, whereby associated osteophytes cause
574 M.T. Nevalainen et al. / Clinical Radiology 71 (2016) 570e575
Figure 3 Follow-up imaging of cervical facet oedema. (a) In the first MR image, moderate BMO with adjacent soft-tissue oedema is seen at the
right C4e5 facet joint (arrow) in a 75-year-old woman. (b) On the follow-up MR image 27 months later, the BMO has resolved completely leaving
only a small amount of fluid signal within the facet joint (arrow).
narrowing of the neural foramen and compression of the
traversing nerve roots; however, in the present analysis, the
ipsilateral neuroforaminal narrowing was not associated
with cervical facet oedema in general or with the symptom-
corresponding BMO cases. Although the evaluation of
neuroforaminal narrowing is highly observer dependent,10
the present results suggest that neuroforaminal narrowing
at the cervical spine does not correlate with facet oedema.
The severity of BMO seen at cervical facets was further
classified as mild, moderate, or severe. BMO has been
shown be a concise marker of musculoskeletal pain in pe-
ripheral joints, such as the knee,14,15 but also elsewhere in
the musculoskeletal system.16 Here the poorest correlation
was observed between severe BMO at the cervical facets
and unilateral neck pain and/or radiculopathy, although
there were no remarkable differences in correlation be-
tween the BMO groups (Table 2). Ultimately, the discrep-
ancy concerning the severity of BMO and corresponding
symptoms might be explained by the small size of the se-
vere BMO subgroup. Nevertheless, the relationship between
unilateral symptoms and ipsilateral BMO overall was sig-
nificant. Additionally, soft-tissue oedema surrounding the
facets clearly increased with the severity of the BMO.
On examination of the follow-up MRI reports for cervical
facet oedema, BMO remained unchanged in 10/27 cases,
decreased in 12/27 cases, and increased in 5/27 cases. In a
similar study on lumbar facet oedema, Friedrich et al. (2007)
reported on nine patients with follow-up MRI examina-
tions, and found that BMO remained stable in 3/9 patients,
decreased in 4/9 patients, and increased in 2/9 patients.13 A
resemblance to the present results can, therefore, be seen,
but both studies had rather small follow-up groups, and
therefore, no drastic conclusions can be drawn. In addition
to these 27 follow-up cases, nine patients had prior normal
cervical facets at MRI, but later imaging showing BMO at the
facets. Accordingly, it could be suggested that BMO at the
cervical facets is a phenomenon that comes and goes mostly
related to mechanical stress; however, in the end the pre-
sent clinical data were insufficient to make any correlation
to the neck pain or radiculopathy with follow-up imaging.
There were limitations to the present study, including
those inherent in any retrospective study. Correlation of the
site of facet oedema with the precise radicular distribution
of the symptoms was not performed because the clinical
data were mostly inconclusive to define a specific cervical
spine level. It was not possible to prove that the facet joint
was the source of pain in patients, where facet oedema
correlated with the side of symptoms. Other cervical pa-
thology, such as cervical disc disease, osteophytic spurring,
vertebral listhesis, and spinal cord and nerve root
compression, are other potential sources of pain and con-
founding factors in this patient group. In addition, only the
cervical level affected with BMO was further analysed for
neuroforaminal narrowing. Follow-up imaging was only
available for a small group of patients with insufficient
clinical data to draw any clinical conclusions.
Peri-articular marrow T2/STIR hyperintense signal is
commonly found in association with degenerative change
at other joints throughout the body. In all of the positive
cases in the present study, there was concomitant degen-
erative change. Infection and inflammatory arthropathy
may also be considered in cases of peri-articular T2/STIR
hyperintense signal abnormality; however, no evidence of
progressive infection or inflammatory arthropathy
requiring follow-up imaging or intervention was found in
any of the positive cases in the present study.
In conclusion, the prevalence of cervical facet oedema
was 9% at MRI performed for evaluation of neck pain and/or
upper extremity radiculopathy. There was a significant
correlation between cervical facet oedema and ipsilateral
neck pain and/or radicular symptoms, which did not,
however, significantly increase with severity of the BMO at
the facets. Neuroforaminal narrowing was not associated
with cervical facet oedema. Moreover, in the follow-up
 M.T. Nevalainen et al. / Clinical Radiology 71 (2016) 570e575
imaging, facet oedema was most likely to remain un-
changed or decrease.
References
1. Fejer R, Kyvik KO, Hartvigsen J. The prevalence of neck pain in the world
population: a systematic critical review of the literature. Eur Spine J 2006
Jun;15(6):834e48.
2. Hoy D, March L, Woolf A, et al. The global burden of neck pain: estimates
from the global burden of disease 2010 study. Ann Rheum Dis 2014
Jul;73(7):1309e15.
3. Cohen S. Epidemiology, diagnosis, and treatment of neck pain. Mayo Clin
Proc 2015 Feb;90(2):284e99.
4. Aprill C. Cervical zygapophyseal joint pain patterns II: a clinical evalu-
ation. Spine 1990;15(6):458e61.
5. Fukui S, Ohseto K, Shiotani M, et al. Referred pain distribution of the
cervical zygapophyseal joints and cervical dorsal rami. Pain 1996
Nov;68(1):79e83.
6. Kotsenas AL. Imaging of posterior element axial pain generators: facet
joints, pedicles, spinous processes, sacroiliac joints, and transitional
segments. Radiol Clin N Am 2012;50(4):705e30.
7. Lee MJ, Riew KD. The prevalence cervical facet arthrosis: an osseous
study in a cadveric population. Spine J 2009 Sep;9(9):711e4.
8. Corey DL, Comeau D. Cervical radiculopathy. Med Clin North Am 2014
Jul;98(4):791e9.
575
9. Radhakrishnan K, Litchy WJ, O’Fallon WM, et al. Epidemiology of cervical
radiculopathy. A population-based study from Rochester, Minnesota,
1976 through 1990. Brain 1994 Apr;117(Pt 2):325e35.
10. Kuijper B, Tans JT, van der Kallen BF, et al. Root compression on MRI
compared with clinical findings in patients with recent onset cervical
radiculopathy. J Neurol Neurosurg Psychiatry 2011 May;82(5):561e3.
11. Zanetti M, Bruder E, Romero J, et al. Bone marrow oedema pattern in
osteoarthritic knees: correlation between MR imaging and histologic
findings. Radiology 2000 Jun;215(3):835e40.
12. Gellhorn A, Katz J, Suri P. Osteoarthritis of the spine: the facet joints. Nat
Rev Rheumatol 2013 Apr;9(4):216e24.
13. Friedrich KM, Nemec S, Peloschek P, et al. The prevalence of lumbar facet
joint oedema in patients with low back pain. Skelet Radiol
2007;36(8):755e60.
14. Starr A, Wessely M, Albastaki U, et al. Bone marrow oedema: patho-
physiology, differential diagnosis, and imaging. Acta Radiol 2008
Sep;49(7):771e86.
15. Sharkey P, Cohen S, Leinberry C, et al. Subchondral bone marrow lesions
associated with knee osteoarthritis. Am J Orthop (Belle Mead NJ) 2012
Sep;41(9):413e7.
16. Meizer R, Radda C, Stolz G, et al. MRI-controlled analysis of 104 patients
with painful bone marrow oedema in different joint localizations
treated with the prostacyclin analogue iloprost. Wien Klin Wochenschr
2005 Apr;117(7e8):278e86.