Keywords
bone marrow edema - periarticular fat deposition - sacroiliitis - subchondral sclerosis
- T2 Dixon
Introduction
Sacroiliac joint is involved in multiple rheumatic and nonrheumatic disorders. Seronegative
spondyloarthropathies (SpA) are the most common cause of sacroiliitis, while other
causes are infection, gout, enteropathic, pyogenic, and rheumatoid arthritis.[1]
[2]
[3]
[4] Clinical diagnosis of sacroiliitis is difficult because patients have localized
or referred pain, usually inferior to the posterior superior iliac spine, mimicking
multiple other causes of back pain.[5]
Radiograph is sensitive only to chronic disease with altered joint space and subchondral
bone changes. Magnetic resonance imaging (MRI) does better evaluation of joint anatomy,
marrow, cartilage, subchondral bone, surrounding ligaments, and capsule and is superior
in early diagnosis of acute sacroiliitis.[6]
[7] It is included in the new criteria defined by the Assessment of Spondylo Arthritis
International Society, according to which acute sacroiliitis attributable to SpA can
be visualized as bone marrow edema (BME)/osteitis, synovitis, enthesitis, or capsulitis.
Presence of BME/osteitis on MRI is considered essential for diagnosis. Chronic cases
are defined with subchondral sclerosis, periarticular fat deposition, articular erosions,
and bony ankylosis.[8]
[9]
[10] MRI also helps in assessing disease activity and monitoring treatment response in
cases of sacroiliitis.[2]
Conventional MRI sequences of sacroiliac joints are T1-weighted (T1W) and fat-saturated
T2-weighted (T2W-FS)/STIR images. Short tau inversion recovery (STIR) hyperintensities
can be confirmed with contrast-enhanced fat-saturated T1-weighted (T1W-FS-PC) images
to detect osteitis.[2]
[10] STIR is usually sufficient for diagnosis of BME. However, in when STIR is inconclusive
or to exclude possible artifacts, T1W-FS-PC sequence acts as a definitive method.[11]
Fat suppression techniques like chemical shift imaging and related Dixon-based approaches
are used for detection and characterization of liver, renal, adrenal, and focal bone
marrow lesions.[12] T2W multipoint Dixon technique provides four sets of post processed and automatically
reconstructed images.[13]
[14] This Dixon sequence has a better and higher signal-to-noise ratio (SNR) in a reasonable
time with homogeneous fat suppression.[13] Dixon may provide good tissue contrast without time delay compared with the conventional
sequences used in detection and differentiation of sacroiliac pathologies.[12]
[15]
Good-quality imaging and reliable reporting play an important role in detecting the
appropriate candidates with axial disease who would benefit from biologics.[16] This study evaluates the use of Dixon as a single sequence for diagnosis of sacroiliitis
when compared with the conventional protocol of multiple sequences.
Method
Study Population and Study Design
This study was conducted in the Department of Radiology, Kovai Medical Center and
Hospital, Coimbatore, after getting approval from the Institutional Ethics Committee
and Institutional Scientific and Research Committee. Written informed consent was
taken from all patients. All adult patients older than 18 years of age who had undergone
MRI examination of lumbosacral spine and sacroiliac joints in our hospital during
a period of August 2019 to August 2020 were included. The patients with general contraindications
for MRI, patient motion during acquisition, and contraindication for MRI contrast
injection were excluded from the study.
Out of the 1,325 adult patients who underwent MRI examination of lumbosacral spine
and sacroiliac joints during the study period, 37 patients had evidence of sacroiliitis
in conventional MRI sequences and were considered for further evaluation. A total
of 63 sacroiliac joints (26 patients had bilateral involvement) were studied.
MRI Technique
MRI was done either using a Siemens MAGNETOM Skyra 3T MRI or Philips Ingenia 1.5T
MRI with dedicated phased array coils ([Table 1]). Thirty-two patients underwent scans in 1.5 T system and 5 patients in 3T system.
The acquired sequences in the routine protocol were coronal oblique STIR sequences
and T1W turbo spin echo sequences. A T2W multipoint Dixon sequence in coronal oblique
plane was added to the protocol. After the administration of gadolinium, T1W-FS-PC
sequence was performed in the coronal oblique plane if hyperintense signal was noted
in IR sequence.
Table 1
MRI parameters in 1.5T and 3T scanner systems
|
Scanner
|
STIR
|
T1W
|
T1W-FS-PC
|
T2-Dixon
|
|
TR (ms)
|
1.5 T
|
2500–4000
|
400–600
|
400–750
|
3000–5000
|
|
3 T
|
370
|
700
|
574
|
4230
|
|
TE (ms)
|
1.5T
|
55
|
8
|
8
|
80
|
|
3 T
|
36
|
9
|
9
|
81
|
|
Turbo factor
|
1.5 T
|
16
|
5
|
5
|
18
|
|
3 T
|
16
|
5
|
5
|
15
|
|
Slice thickness (mm)
|
1.5 T
|
3.5
|
3.5
|
3.5
|
3.5
|
|
3 T
|
3
|
3
|
3
|
3
|
|
Intersection gap (mm)
|
1.5 T
|
0.35
|
0.35
|
0.35
|
0.35
|
|
3 T
|
0.3
|
0.3
|
0.3
|
0.3
|
Abbreviations: MRI, magnetic resonance imaging; STIR, short tau inversion recovery;
T1W-FS-PC, contrast-enhanced fat-saturated T1-weighted; TR, repetition time; TE, time
to echo.
Reference for normal bone marrow signal was taken as the sacral interforaminal bone
marrow signal. Affected bone marrow areas are typically periarticular. For all sequences,
filed of view was 220 mm.
Image Analysis
Qualitative Analysis
Active inflammation of SI joint was diagnosed if BME was seen in STIR images and osteitis
seen in T1W-FS-PC images. Chronic inflammation was diagnosed if subchondral sclerosis,
periarticular fat deposition, ankylosis, and articular erosions were seen in T1W sequences.
In T2W multipoint Dixon sequence images, active inflammation was looked for in water-only
images and chronic inflammation in in-phase (IP), opposed-phase (OP), and fat-only
(FO) images.
Quantitative Analyses
For all patients, contrast–noise ratios (CNRs) of active subchondral bone marrow lesions,
subchondral sclerosis, and periarticular fat deposition were calculated using:
CNR = [Mean L– Mean BM] / SD air,
-
Mean L: mean signal intensity of the largest lesion in one sacroiliac joint,
-
Mean BM: mean signal intensity of the normal-appearing bone marrow near the lesion,
-
SD air: Standard deviation (SD) of the signal intensity of the airspace free of artifact
and close to the site of previously obtained signal-intensity measurements.[2]
For the same active lesions, CNR was calculated on STIR images, T1W-FS-PC images,
and water-only T2W multipoint Dixon images. For the same chronic lesions, CNR was
calculated on T1W images, IP, OP, and FO T2W multipoint Dixon images. The largest
lesion in each sacroiliac joint of each patient was evaluated. A same-sized round
region of interest was used in all sequences for measuring signal intensity.
Statistical Analysis
Statistical Package for Social Sciences (SPSS) version 20 was employed to analyze
data. Continuous data was analyzed for its mean, median, and SD. Categorical variable
was analyzed using chi-squared test and “p” value of ≤ 0.05 was considered as statistically significant. Data was presented
as mean ± SD for continuous variables and as percentages for categorical variables.
The mean and SD of the CNR value for each prognostic factor were calculated. The analysis
was done to test if there is any statistically significant difference in CNR values
between groups. Paired t-test for independent sample was used to study the difference in CNR between two groups.
Chi-squared analysis was used to test the significant association in qualitative analysis
of different sequences.
Results
A total of 37 patients (20 males and 17 females) and 63 joints (26 patients had bilateral,
7 had right, and 4 had left-sided joint involvement) were studied. The age of the
patients ranged from 19 to 75 years (mean: 38.1 ± 11.8 years). Among these 63 cases
of sacroiliitis, 14 (22.2%) were in acute stage, 29 (46.1%) were in chronic stage,
and 20 cases (31.7%) were in acute on chronic stage. So, a total of 34 SI joints in
acute stage and 49 SI joints in chronic stage were evaluated. Qualitative analysis
was done with visual assessment and quantitative analysis was done with CNR of the
largest lesion in each joint.
Qualitative Analysis
Acute Sacroiliitis
The signal changes consistent with BME/osteitis were visually assessed with STIR,
T1W-FS-PC images, and WO image set of Dixon sequence ([Fig. 1A–C] and [Fig. 2A–C]) and graded as mild and moderate signal changes. All signal changes noted in STIR
and T1W-FS-PC images were also noted in WO images, suggestive of 100% agreement. The
chi-squared test to assess the association between WO and STIR image sets and WO and
T1W-FS-PC image sets yielded a p-value of < 0.001, suggestive of significant strong association between the image
sets in both the groups.
Fig. 1 Oblique coronal sections of short tau inversion recovery (A), fat-saturated T1 postcontrast (B), and water only images (C) of bone marrow edema/osteitis in a case of bilateral acute sacroiliitis.
Fig. 2 Oblique coronal sections of short tau inversion recovery (A), fat-saturated T1 postcontrast (B), and water only images (C) of a case of acute infective left sacroiliitis showing collection and bone marrow
edema/osteitis.
Chronic Sacroiliitis
The signal changes of bone marrow consistent with subchondral sclerosis, periarticular
fat deposition, articular erosions, and ankylosis were visually assessed with T1W
images, IP, OP, and FO T2W multipoint Dixon images and graded as mild, moderate, and
severe changes. All cases detected in T1W images were noted in IP and FO images, suggestive
of 100% agreement. No new cases were diagnosed with Dixon sequences.
A total of 44 cases in T1W, IP, and FO images and 38 cases in OP images demonstrated
subchondral sclerosis ([Fig. 3A–D]). On assessing the association of the above three image sets of Dixon sequence with
T1W images, the IP image sets showed a p-value <0.001, suggestive of strong association. The FO and OP images showed no statistically
significant association with T1W images in detection of subchondral sclerosis.
Fig. 3 Oblique coronal sections of T1-weighted (A), in-phase (B), opposed-phase (C), and fat only images (D) of subchondral sclerosis in bilateral chronic sacroiliitis.
A total of 41cases demonstrated subchondral fat deposition in T1W, IP, OP, and FO
images ([Fig. 4A–D]). On assessing the association of the above three image sets of Dixon sequence with
T1W images, all the three image sets showed a p-value of <0.001, suggestive of strong association.
Fig. 4 Oblique coronal sections of T1-weighted (A), in-phase (B), opposed-phase (C), and fat only images (D) of periarticular fat deposition in bilateral chronic sacroiliitis.
A total of six cases demonstrated ankylosis in T1W, IP, OP, and FO images ([Fig. 5A–D]). On assessing the association of the above three image sets of Dixon sequence with
T1W images, all the three image sets showed a p-value of > 0.05, suggestive of no association.
Fig. 5 Oblique coronal sections of T1-weighted (A), in-phase (B), opposed-phase (C), and fat only images (D) of ankylosis in bilateral chronic sacroiliitis.
A total of 44 cases demonstrated cortical erosions in T1W, IP, and FO images and 24
cases in OP images ([Fig. 6A–D]). On assessing the association of the above three image sets of Dixon sequence with
T1W images, IP and FO sets showed a p-value <0.001, suggestive of significant strong association.
Fig. 6 Oblique coronal sections of T1-weighted (A), in-phase (B), opposed-phase (C), and fat only images (D) of erosions in bilateral chronic sacroiliitis.
Quantitative Analysis
Acute Sacroiliitis
On evaluating 34 cases of acute sacroiliitis quantitatively, the mean CNR of largest
lesions was highest for WO images (296.35; SD: 208.28) compared with STIR and T1W-FS-PC
sequences ([Fig. 7]). On applying paired t-test, statistically significant difference was noted between WO and STIR image sets
(p-value < 0.001) and between WO and T1W-FS-PC image sets (p-value = 0.002) ([Table 2]).
Fig. 7 Distribution of mean values of contrast–noise ratio of bone marrow edema/osteitis
in acute cases. STIR, short tau inversion recovery.
Table 2
Statistical analysis of CNR of BME/osteitis in acute sacroiliitis
|
Acute sacroiliitis
|
Mean
|
SD
|
p-Value
|
|
STIR
|
187.26
|
108.31
|
< 0.001
|
|
WO
|
296.35
|
208.28
|
|
T1W-FS-PC
|
218.50
|
145.59
|
0.002
|
|
WO
|
296.35
|
208.28
|
Abbreviations: BME, bone marrow edema; CNR, contrast–noise ratio; SD, standard deviation;
STIR, short tau inversion recovery; T1W-FS-PC, contrast-enhanced fat-saturated T1-weighted.
Chronic Sacroiliitis
Subchondral Sclerosis
On evaluating 44 cases of chronic sacroiliitis with subchondral sclerosis quantitatively,
the mean CNR of largest lesions was highest for T1W images (186.09; SD: 96.96) compared
with IP, OP, and FO sets of T2 Dixon sequences ([Table 3]). On applying paired t-test, statistically significant difference was noted between T1W and OP image set
and T1W and FO image sets (p-value < 0.001). A p-value was not statistically significant (>0.05) when T1W image was compared with
IP images.
Table 3
Statistical analysis of CNR of subchondral sclerosis in chronic sacroiliitis
|
Subchondral sclerosis
|
Mean
|
SD
|
p-Value
|
|
T1W
|
186.09
|
96.96
|
0.107
|
|
IP
|
162.32
|
109.22
|
|
T1W
|
189.50
|
89.65
|
< 0.001
|
|
OP
|
50.73
|
34.36
|
|
T1W
|
186.09
|
96.96
|
< 0.001
|
|
FO
|
122.27
|
63.48
|
Abbreviations: CNR, contrast–noise ratio; FO, fat only; IP, in-phase; OP, opposed-phase;
SD, standard deviation; T1W, T1-weighted.
Fat Deposition
On evaluating 41 cases of chronic sacroiliitis with fat deposition quantitatively,
the mean CNR of largest lesions was highest for OP images (279.22; SD: 188.40) compared
with T1W, IP, and FO images ([Table 4]). On applying paired t-test, statistically significant difference was noted between T1W and OP image set
(p-value: < 0.001) and T1W and FO image set (p-value = 0.001). A p-value was not statistically significant (p-value: > 0.05) when T1W image was compared with IP images.
Table 4
Statistical analysis of CNR of fat deposition in chronic sacroiliitis
|
Fat deposition
|
Mean
|
SD
|
p-Value
|
|
T1W
|
153.41
|
101.61
|
0.338
|
|
IP
|
141.71
|
121.42
|
|
T1W
|
153.41
|
101.61
|
< 0.001
|
|
OP
|
279.22
|
188.40
|
|
T1W
|
153.41
|
101.61
|
0.001
|
|
FO
|
193.44
|
95.96
|
Abbreviations: CNR, contrast–noise ratio; FO, fat only; IP, in-phase; OP, opposed-phase;
SD, standard deviation; T1W, T1-weighted.
On comparing the scan times in 3T Siemens MRI, collective scan time for routine STIR
(53 seconds), T1W imaging (1 minute, 41seconds), and T1W-FS-PC image (1 minute, 5 seconds)
is 3 minutes, 39 seconds. The time for T2 multipoint Dixon sequence is 2 minutes,
37 seconds.
In 1.5T Philips MRI, collective scan time for routine STIR (1 minute, 45 seconds),
T1W imaging (2 minutes, 19 seconds), and T1W-FS-PC (3 minutes, 34 seconds) imaging
is 7 minutes, 38 seconds. The time for T2 multipoint Dixon sequence is 3 minutes,
35 seconds.
Discussion
The use of Dixon sequences in various clinical settings has shown to provide a better
fat suppression and image quality when compared with the conventional methods. On
the basis of this rationale, Dixon sequence is expected to be equal or superior to
the conventional imaging of sacroiliac joint. This could reduce the scanning time
by decreasing a multisequence protocol to a single Dixon T2W sequence. It is also
expected to improve the image quality leading to increased diagnostic confidence.
The aim of the study was to assess the profile of T2W multipoint Dixon sequence and
conventional sequences in MRI of sacroiliac joints for the diagnosis of active and
chronic sacroiliitis.
Only very few studies have been conducted regarding the role of Dixon in the evaluation
of sacroiliac joint. The study conducted by Özgen in 2017 concluded that T2W multipoint
Dixon sequence was superior to conventional MRI sequences in depicting diagnostic
signs of active and chronic sacroiliitis and therefore may be used as a single sequence.[2]
In our study, the bone marrow lesions in acute sacroiliitis were equally appreciated
in STIR, T1W-FS-PC, and WO images. The WO images were superior to STIR and T1W-FS-PC
images quantitatively for the detection of BME/osteitis. However, the administration
of contrast did not add to the diagnosis of any new cases of sacroiliitis.
In chronic sacroiliitis, subchondral sclerosis was demonstrated adequately in IP sequences
and was comparable with T1W sequences. The diagnosis of subchondral sclerosis was
difficult in FO and OP sequences. No statistically significant difference was noted
between T1W and IP images on quantitative analysis though mean CNR was higher for
T1W images. T1W sequence was superior to FO and OP sequences and was comparable to
IP sequences.
Periarticular fat deposition was demonstrated in all the three Dixon sequences. The
signal changes in OP images were qualitatively comparable to or better than the T1W
sequences. The OP images followed by FO images were superior in the assessment of
periarticular fat deposition compared with routine sequences.
In assessing ankylosis qualitatively, no statistically significant association existed
between T1W sequence and any of the imaging sets of Dixon sequence. Concerning cortical
erosions, all lesions noted in the conventional T1W images were redemonstrated in
IP image sets. Also, the lesions in IP and FO image sets were comparable to T1W images.
However, the diagnosis of these conditions was difficult in OP images.
Therefore, in our study, using T2W multipoint Dixon sequence for analysis of sacroiliitis,
WO images were superior to the conventional imaging modality in detecting acute bone
marrow lesions; IP images were comparable to conventional T1W sequences in detecting
subchondral sclerosis and OP images followed by FO images were superior to T1W images
in detecting periarticular fat deposition. Detection of cortical erosions by IP and
FO image sets were comparable to T1W images. Diagnosis of ankylosis was difficult
with the Dixon sequences.
According to the study conducted by Özgen in 2017, who studied 34 lesions among 73
patients, a similar outcome was obtained. However, the T2W-FS sequence was considered
superior to postcontrast images in acute stage. He showed that use of IP images could
result in a statistically significant difference in detecting subchondral sclerosis.
T1W images had the minimum CNR considering periarticular fat deposition and statistically
significant difference was noted between T1W and OP images only.[2]
The advantage of Dixon is that due to their insensitivity to local magnetic field
inhomogeneities, it provides homogenous fat suppression compared with other fat saturation
techniques. These techniques can also be used for fat quantification. Dixon techniques
can be implemented in both gradient echo and spin-echo sequences, with some technical
adaptation.[13] However, metallic artifacts due to larger prosthesis result in low-quality images
in case of Dixon technique. In this setting, STIR remains the preferred technique.[12]
[13] However, it has a low SNR compared with Dixon sequences and cannot be used as a
postcontrast sequence. Hence, Dixon is a preferred in knee with small metallic material,
in ankle or for spinal arthrodesis due to its higher SNR. It is also used when fat
suppression is needed in postcontrast images as in inflammatory/infectious changes
in the proximity of metallic implants.[13]
Dixon sequence is widely used in many fields in recent history. The Dixon sequences
provide both better image quality and increased CNR when compared with the routine
fat suppression sequences used.[17] Due to the improved CNR and decreased acquisition time with advantage that it can
be used in both T1W and T2W sequences, it is considered equal to or superior to different
conventional modalities and in imaging of neck, breast, abdomen, and female pelvic
conditions.[13]
[17]
[18]
[19]
[20]
[21] It is also studied in recent fields of renal dynamic contrast-enhanced MRI and cardiac
imaging.[22]
[23] Dixon sequence has many applications in musculoskeletal imaging where the bone marrow
replacement lesions and fractures can be detected with increased accuracy.[12]
[24]
[25]
[26]
Most of these studies suggest that Dixon as a single sequence can replace the standard
combination protocol of morphologic sequences followed in these areas.
Conclusion
In detecting acute sacroiliitis, WO images of the T2 multipoint Dixon sequence are
superior to the conventional MRI sequences. Thus, Dixon can be used as a single sequence
to replace the multiple sequences used in conventional imaging protocol of sacroiliac
joints due to higher image quality and reduced scan time in case of acute sacroiliitis.
In detecting chronic sacroiliitis, OP images followed by FO images are superior to
T1W images in detecting periarticular fat deposition and IP images are comparable
to T1W sequences in detecting subchondral sclerosis. Detection of cortical erosions
by IP and FO image sets is comparable to T1W images. Diagnosis of ankylosis is difficult
with the Dixon sequences. Hence, Dixon can be used as an additional sequence to the
existing standard sequences in case of chronic sacroiliitis, especially to increase
the confidence of the radiologist and accuracy in diagnosis.
Limitations
Since the sample size is small, larger studies are needed for better analysis of the
possibilities of Dixon sequences. The data involves scans done in both 1.5T and 3T
MRIs from different vendors. The diagnosis of ankylosis with Dixon technique showed
no significant association with the conventional sequence, likely due to limited number
of patients studied with ankylosis. Both the conventional and Dixon sequences were
analyzed only by one observer and was done at the same time, which may cause bias
affecting the outcome. Sacroiliitis due to multiple causes like inflammatory and infective
etiologies is included in the study group. The reference for diagnosis of sacroiliitis
was taken as the conventional imaging protocol and was not proved by laboratory findings
or follow-ups. Follow-up of the cases with Dixon sequence is not done to evaluate
the serial change in signal characters, and thereby its scope in assessing treatment
response is suboptimally assessed.
Recommendations
Dedicated studies with higher patient population are needed to support the use of
Dixon sequence as the primary sequence for screening sacroiliac joints. Also, more
patients with ankylosis need to be studied to assess the role of Dixon sequence in
its diagnosis. Proper administration of matching and blinding techniques is needed
in the future studies to avoid errors and bias. Role of Dixon technique in follow-up
cases also needs to be studied in detail.
