MRI findings of treated bacterial septic arthritis
- First Online:
- Cite this article as:
- Bierry, G., Huang, A.J., Chang, C.Y. et al. Skeletal Radiol (2012) 41: 1509. doi:10.1007/s00256-012-1397-2
- 747 Views
The purpose of this study was to report the MRI findings that can be encountered in successfully treated bacterial septic arthritis.
Materials and methods
The study included 12 patients (8 male and 4 female; mean age 38 years, range 9–85) with 13 proven cases of bacterial septic arthritis. The joints involved were hip (n = 3), knee (n = 3), shoulder (n = 2), sacroiliac (n = 2), ankle (n = 1), wrist (n = 1), and elbow (n = 1). MRI examinations following surgical debridement and at initiation of antibiotic therapy and after successful treatment were compared for changes in effusion, synovium, bone, and periarticular soft tissues. Imaging findings were correlated with microbiological and clinical findings.
Joint effusions were present in all joints at baseline and regressed significantly at follow-up MRI (p = 0.001). Abscesses were present in 5 cases (38 %), and their sizes decreased significantly at follow-up (p = 0.001). Synovial enhancement and thickening were observed in all joints at both baseline and follow-up MRI. Myositis/cellulitis was present in 10 cases (77 %) at baseline and in 8 cases (62 %) at follow-up MRI. Bone marrow edema was present in 10 joints (77 %) at baseline and persisted in 8 joints (62 %). Bone erosions were found in 8 joints (62 %) and persisted at follow-up MRI in all cases.
The sizes of joint effusions and abscesses appear to be the factors with the most potential for monitoring therapy for septic arthritis, since both decreased significantly following successful treatment. Synovial thickening and enhancement, periarticular myositis/cellulitis, and bone marrow edema can persist even after resolution of the infection.
Septic arthritis is a medical and surgical emergency that has an impact on both the functional prognosis of the involved joint as well as the overall prognosis for the patient. Treatment monitoring of septic arthritis is based on clinical findings (regression of inflammatory symptoms) and microbiological or laboratory findings (white blood cell [WBC] count, C-reactive protein [CRP]) . Septic arthritis only improves slowly under antibiotic therapy, and determining whether abnormal clinical findings (pain, swelling) are related to a persistent infection or are manifestations of normal recovery can be challenging. In this setting, the sensitivity and predictive values of blood markers such as CRP and WBC are relatively low . Analysis of synovial fluid and determination of synovial WBC count can be helpful in determining response to therapy, but requires repeated invasive arthrocentesis. Furthermore, bacteriological analysis is known to be unreliable while the patient is under antibiotic treatment .
Therefore, a non-invasive imaging technique that could reliably demonstrate the status of joint infection, help to determine the duration of treatment and the timing of its discontinuation, and show the occurrence of relapse would be a valuable monitoring tool. MRI is the imaging modality of choice for the evaluation of synovial disorders, and in septic arthritis, nonspecific synovial thickening and enhancement and joint effusions are generally observed . Despite its high sensitivity, MRI plays only a minor role for the initial diagnosis of septic arthritis, since the first imaging abnormalities appear only after several days of infection . MRI is usually performed to exclude other potential diagnoses, to target surgical debridement, to demonstrate associated osteomyelitis or abscess formation, or to explore deep joints (hip, sacroiliac joint) . MRI findings observed during treatment, such as synovitis and bone marrow edema, may be difficult to interpret. To date, reports of MRI changes in successfully treated septic arthritis in humans remain rare and are focused on the axial skeleton . The purpose of this study was to report MRI findings that can be encountered in successfully treated septic arthritis in different peripheral joints.
Materials and methods
This study was approved by our Institutional Review Board, and informed consent was waived. This study was compliant with the Health Insurance Portability and Accountability Act.
Bacteriologically proven nontuberculous septic arthritis
Available MRI following surgical debridement and at initiation of antibiotic therapy (baseline MRI) and follow-up MRI at the time of successful treatment, including T1-weighted, fat-suppressed T2-weighted, and gadolinium-enhanced fat-suppressed T1-weighted images
Available clinical and laboratory data at the time of follow-up suggesting successfully treated infection, such as the absence of fever, resolution of joint inflammatory signs (pain, swelling, motion), normalization of WBC and CRP levels or negative aspiration
Available clinical follow-up of at least 3 months after the last MRI that confirmed resolution of infection 
A total of 12 patients with 13 affected joints met these inclusion criteria. The study population consisted of 8 male and 4 female patients with a mean age of 38 years (range, 9–85 years). The joints involved were hip (n = 3), knee (n = 3), shoulder (n = 2), sacroiliac (n = 2), ankle (n = 1), wrist (n = 1), and elbow (n = 1). The baseline MRI examination was performed 13.5 days (range, 0–40) after antibiotic therapy initiation; the follow-up examinations were performed 52.1 days (range, 19–155) after the baseline examination.
Image acquisition and analysis
Magnetic resonance imaging was performed on 1.5-T or 3T scanners (GE Medical Systems, Milwaukee, WI, USA, or Siemens Medical Solutions, Erlangen, Germany) using routine imaging protocols that included at least T1-weighted (450–600/10–14 repetition time [TR] ms/echo time [TE] ms), fat-suppressed T2-weighted (3,300–4.500/40-80), and fat-suppressed T1-weighted pulse sequences (450–600/10–14) after IV injection of 0.1 mmol/kg of gadopentetate dimeglumine (Magnevist, Bayer HealthCare, Wayne, NJ, USA). The field of view ranged from 10 × 10 to 40 × 40 cm and the matrix from 256 × 256 to 512 × 512. Dedicated extremity coils were used for the wrist, shoulder, and knee, and body surface coils were used for the hip and sacroiliac joints.
Both qualitative and quantitative image analysis was performed. For the qualitative image analysis, two musculoskeletal radiologists (radiologist 1 [MAB], with 11 years of experience; radiologist 2 [GB], with 5 years of experience) evaluated all MRI examinations in consensus. Examinations were reviewed in a random order to avoid memory effect, and readers were blinded to the status (baseline, follow-up) of MRI examinations. The following findings were evaluated: bone abnormalities (erosions and marrow edema), joint effusion, synovial enhancement, extra-articular abscess, and periarticular soft tissue/muscle abnormalities. Bone erosions were defined as well-demarcated osseous lesions with disruption of the normal cortical contour . Bone marrow edema was considered to be indicated by an ill-defined lesion within the trabecular bone with signal characteristics consistent with increased water content (low T1 signal, high T2 signal, enhancement after gadolinium administration) [9, 10]. Synovial enhancement on gadolinium-enhanced fat-suppressed T1-weighted images was scored 0 for no enhancement, 1 for mild, 2 for moderate. and 3 for marked enhancement . Joint effusion on T2-weighted images was scored 0 for no effusion, 1 for small, 2 for medium, and 3 for large effusion with capsular distension . When infection presented with extra-articular extension, the presence of abnormal soft tissue/muscle signal and/or abscess was noted. An ill-defined area of diffuse enhancement was indicative of myositis/cellulitis, and an area of rim enhancement with central non-enhancement was considered an abscess .
On the baseline examination, synovial characteristics were evaluated using the slice and imaging plane with the maximum synovial thickness. This also served as the reference slice for follow-up measurements. Joint effusion and abscess size were evaluated on the slice with the greatest amount of fluid at baseline. Joint effusion, synovial thickness, and abscess size (maximal short axis dimension) were measured using the picture archival and communication system (PACS) workstation calipers. Synovial signal was measured using a circular region of interest (ROI) of a mean size of 10 mm2 .The ROI was placed on the area with the highest signal intensity on visual analysis. Signal-to-noise ratio (SNR) was calculated by dividing the measured signal of the synovium by the standard deviation of the noise measured using a ROI of 20 mm2 placed in the air next to the joint.
Statistical analysis was performed using JMP software (version 8, 2010; SAS, Cary, NC, USA). Changes in synovial thickening, synovial signal, joint effusion, and abscess size were compared using the Wilcoxon rank test for paired values. Changes in bone signal, bone erosion, and periarticular soft tissue abnormalities (cellulitis/myositis and abscess) were compared using the McNemar test. p values less than 0.05 were considered statistically significant.
ATB, debridement, drainage
Group B streptococcus
Group B streptococcus
ATB, debridement, drainage
At the time of the follow-up MR examination, all patients presented with clinical improvement, including absence of general (fever, malaise) or local (redness, swelling) signs of infection. Three patients (25 %) reported nonspecific persistent pain. WBC and CRP levels reverted to normal in all patients and one patient had repeated joint aspiration that was negative for infection.
Magnetic resonance imaging findings at baseline and follow-up
MR imaging findings
Baseline (n = 13)
Follow-up (n = 13)
6 (46 %)
2 (15 %)
7 (54 %)
9 (70 %)
2 (15 %)
4 (31 %)
8 (62 %)
1 (8 %)
8 (62 %)
5 (38 %)
Bone marrow edema
10 (77 %)
8 (62 %)
8 (62 %)
8 (62 %)
10 (77 %)
8 (62 %)
5 (38 %)
1 (8 %)
Effusions were observed in all joints at baseline (grade 3, n = 4, grade 2, n = 8; grade 1, n = 1; Figs. 2, 3, 4 and 5). At follow-up, 8 joints had mild persistent effusion (grade 1, n = 8; p = 0.001; Figs. 4 and 5) and in 5 joints the effusions had resolved. Quantitative analysis confirmed the visual evaluation, showing a significant reduction of effusion, from 8.8 ± 8 mm at baseline to 2.1 ± 1.7 mm at follow-up (change = −76 %, p = 0.002).
Bone marrow edema was present in 10 joints at baseline and in 8 joints at follow-up (Figs. 2, 3, 4 and 5; p = 0.5). Bone erosions were observed in 8 joints at baseline and persisted at follow-up in all cases (p = 1; Fig. 3).
Periarticular soft tissue abnormalities were present in 11 joints (85 %) at baseline (cellulitis/myositis in 6 joints, cellulitis/myositis and abscess in 4 joints, abscess in 1 joint). At follow-up, 8 joints had persistent areas of myositis/cellulitis (p = 1; Figs. 2, 3, 4 and 5). Of the 5 abscesses at baseline, 4 had resolved (Fig. 5), and 1 had almost completely resolved at follow-up. Two of the 5 abscesses had been percutaneously drained without recurrence; the 3 other abscesses resolved under antibiotic therapy alone. Although the small number of abscesses did not show a significant difference in the qualitative analysis (p = 0.125), a significant overall reduction of abscess size (from 13.8 ± 5.2 mm to 1 ± 2.2 mm, change = − 92 %, p = 0.001) was found on quantitative analysis.
This study demonstrates that MRI examinations of bacterial septic arthritis performed after successful therapy can show persistent synovial thickening and enhancement, and bone marrow and periarticular soft tissue edema, while joint effusions and abscesses significantly decrease after successful treatment.
The treatment of acute bacterial arthritis requires antibiotic therapy and removal of purulent material from the joint, either surgically (by arthroscopy or arthrotomy and washout) or through percutaneous needle aspiration . Monitoring treatment response of musculoskeletal infections is based on clinical evaluation directed toward local (tenderness, swelling, restricted motion of the affected limb) or general (fever, malaise) symptoms . Serum markers, such as CRP and WBC are recognized as lacking sensitivity and predictive value in the setting of persistent inflammation . In effectively managed infections, CRP classically normalizes in 10 days, but elevated levels can also be observed, especially in prosthetic joints . Synovial fluid analysis (biochemical, bacteriological) is suboptimal for monitoring infection because it poorly reflects the joint status and requires repeated arthrocentesis. Synovial CRP and WBC count may remain elevated even in cases where infection is known to have been eradicated . Finally, conventional bacteriological analysis (Gram staining, culture) is unable to detect persistent infection during antibiotic therapy . MRI is often performed to monitor response to therapy. However, MRI findings are often confusing and can show persistent abnormalities even in cases of successfully treated infection. Therefore, awareness of persistent MRI findings of successfully treated infection might help to monitor therapy more effectively and might prevent continued and unnecessary therapy, despite treatment success.
Classical MRI findings of septic arthritis include joint effusion, synovial thickening, synovial enhancement, bone erosions, and bone marrow and soft tissue edema . In our study, we observed these findings in all patients at baseline. On follow-up MRI examinations, when the joints were considered successfully treated, joint effusions had significantly regressed, and in many cases, completely resolved. However, synovial enhancement and thickening persisted. Persistent synovial thickening might reflect persistent inflammatory activity in the joint not detectable by clinical or laboratory examinations. In addition, MRI findings often lag behind the clinical course and often take longer to resolve completely. The enhancement of fibrous scarring after contrast medium administration has been extensively investigated in the axial skeleton, but less in the peripheral skeleton. Ross et al. reported that post-surgical epidural fibrous scarring can demonstrate persistent enhancement on immediate and delayed contrast-enhanced images .
Different mechanisms might explain synovial MRI changes in the acute inflammatory phase and during the healing phase. During the acute phase, vasodilatation, angiogenesis, and increased vascular permeability occur, explaining the synovial enhancement, extracellular edema, and joint effusion . During the healing phase, enhancement is due to the sequestration of contrast medium within the enlarged extracellular space that is the scarring fibrous synovium . As vascular permeability is almost normal in this phase, the joint effusion is no longer generated and can be drained by lymphatic vessels. In our study, reduction of joint effusion was present in all cases at follow-up and appears to be a good marker of infection resolution.
We found bone marrow edema in 10 cases (77 %) at the initial examination, which persisted in 8 joints at follow-up. Therefore, persistent bone marrow edema following resolution of infection is likely due to destruction of the articular cartilage and increased mechanical changes. In a study of 80 patients with post-traumatic bone marrow edema of MRI of the knee, Boks et al. showed that regression of abnormalities occurred after a mean time of 42 weeks , suggesting that even after discontinuation of the insult (such as trauma), bone marrow healing is a long-lasting process.
Bone erosions were present in 8 joints (62 %) and persisted in all cases at follow-up. Erosions represent destruction of the bony cortex and subchondral plate. Repair of erosions has been more extensively investigated in inflammatory arthropathies . In patients with rheumatoid arthritis under therapy, repair of erosions can be observed, but such regeneration can take up to 12 months to become visible . This delayed repair extends beyond the longest follow-up time we had in our study, thus explaining the persistence of erosions in our cases. Consequently, the presence or absence of erosions is not a useful marker for successful treatment of septic arthritis.
Periarticular soft tissue abnormalities are common associated findings in septic arthritis and include myositis/cellulitis or abscess formation . We observed soft tissue changes in 11 joints (85 %) at baseline MRI examination. At follow-up, 8 joints showed persistent myositis/cellulitis, while of the 5 abscesses, 4 had completely resolved and 1 had almost completely resolved, making the presence or size of abscess a useful indicator of treatment success.
Our study had several limitations, including bias related to the population selection and data collection. First, we included only patients with bacteriologically proven septic arthritis and thus may have selected a particularly virulent organism or a massive infection. In addition, our patient cohort was relatively small and included various joints. Second, the range of baseline MRI findings was large, which could introduce interpretation bias, since the phase of infection (acute versus healing) influences disease extent and the pathological features of the infection. Third, the interobserver reliability was not assessed. Lastly, we used composite criteria to indicate treatment success, including improvement of clinical findings (absence of fever, joint redness and swelling, and increased range of motion) and at least 3 months of follow-up and improvement of serological findings (normalization of CRP and WBC). However, such composite criteria are used as the standard of reference and have also been used to define patient clinical improvement in cases of spinal infection . No cases with persistent/recurrent infection were evaluated.
In conclusion, our study demonstrated that in successfully treated septic arthritis MRI abnormalities can persist, in particular synovial thickening and enhancement, and bone marrow and periarticular soft tissue edema. Additionally, joint effusion and periarticular abscess size may be useful for monitoring therapy, as these were seen to significantly regress following successful therapy.