“Rice Body”-Like Synovial Particles on Magnetic Resonance Imaging: What Is the Spectrum of Diagnoses?

Abstract

Background:

“Rice body” describes the gross appearance of hyperplastic synovial villi that develop into rice-like fibrinous loose bodies. Rice bodies have historically been associated with tuberculosis and inflammatory arthritis, but no retrospective studies have correlated their magnetic resonance imaging (MRI) appearances to histology or to the underlying diagnoses.

Purpose:

We sought to correlate the appearance on MRI of rice body-like particles with histology and describe the associated diagnoses.

Methods:

We conducted a retrospective cross-sectional study of MRI reports in our institutional imaging database in which a radiologist described “rice bodies.” The search was performed by a keyword search for “rice,” ultimately yielding a total of 100 patients between June 1992 and July 2022. Patients were excluded if their MRI was not available for retrospective review or if there was insufficient follow-up to establish a clear underlying diagnosis. Cases were included if the MRI finding of rice bodies was confirmed by retrospective review of the images and there was clinical follow-up or surgery with histology establishing the underlying diagnosis.

Results:

Of the 52 patients who met inclusion criteria and were enrolled in our retrospective study, 37% (19/52) had prior surgery, 73% (14/19) being arthroplasty. Diagnoses associated with rice body-like particles on MRI were inflammatory arthritis (52%; 27/52), infection (25%; 13/52), synovial chondromatosis (8%; 4/52), osteoarthritis (6%; 3/52), bursitis (6%; 3/52), and adverse local tissue reaction (4%; 2/52). Among 13 infections, only 1 was mycobacterial while 46% (6/13) were staphylococcal and 31% (4/13) were culture-negative. In the presence of arthroplasty, rice body-like particles were associated with infection in 71% (10/14), followed by adverse local tissue reaction (14%; 2/14), 46% (24/52) underwent surgery after MRI, and 45% (10/22) had rice bodies confirmed on histology.

Conclusion:

Our retrospective findings suggest that rice body-like synovial particles on MRI are not specific to tuberculosis and inflammatory arthritis. Other diagnoses may also be considered, particularly nonmycobacterial infection when there is an arthroplasty.

Level of Evidence:

Level IV: retrospective cross-sectional study.

Introduction

Rice bodies are small, filiform structures that can develop within joints, bursae, and tendon sheaths and can be associated with joint pain, stiffness, restricted movement and, rarely, with mechanical symptoms such as locking or catching.^1-3 First described by Riese in 1895, the term “rice bodies” was based on their resemblance to polished grains of white rice on gross inspection; they have also been associated with the histologic appearance of fibrin particles in various states of organization, referred to as “fibrinous loose bodies” (Figure 1).^4-6 Although the underlying cause is not fully understood, it is believed that hyperplastic and inflamed synovial villi detach, resulting in rice-like particles containing necrosis, fibrin, and inflammatory cells (Figure 2). ⁷

Figure 1.

Intraoperative image from a knee arthroscopy shows multiple white ovoid intra-articular bodies that are similar in size and shape, consistent with rice bodies.

Figure 2.

(A) Histologic images at low-power view of rice bodies demonstrates numerous bodies which were resected during a knee arthroscopy and (B) on high-power view, there is a central fibrin core with areas of necrosis (black arrow). At the periphery, there are small lymphocytes (white arrow).

Historically, rice bodies have been attributed to a limited number of clinical entities, including tuberculous infection^8-11 and inflammatory arthropathies.^12,13 Other less common diagnoses have also been associated with rice bodies, such as nontuberculous mycobacterial infection, osteoarthritis, bursitis, and tenosynovitis. In rare cases, no underlying etiology is identified.^3,7,14-18 These associations have been described mainly in case reports, not large cohort studies.

While the diagnosis of rice bodies is usually made on gross inspection, intraarticular particles that appear as fine hypointense monomorphic bodies on T2- or intermediate-weighted sequences may uncommonly be encountered on magnetic resonance imaging (MRI), and radiologists may refer to them as “rice bodies.” Case reports have described this as the MRI appearance of histologically confirmed rice bodies, but no published study has looked at how specific this MRI finding actually is for the histologic diagnosis of rice bodies, nor at the diversity and frequency of underlying diagnoses when these intraarticular particles are seen on MRI.^{7,12,14,15,19} The purpose of this retrospective cross-sectional study was (1) to assess the concordance of rice body-like intraarticular particles seen on MRI with true rice bodies and (2) to describe the range and frequency of diagnoses that are associated with this MRI finding.

Methods

We conducted a retrospective cross-sectional study in compliance with the Health Insurance Portability and Accountability Act and approved by our Institutional Review Board. A search of our institutional imaging report database was performed using the keyword “rice,” with the results filtered to include only MRI exams (Montage Health Solutions/Nuance), with the goal of identifying all MRI reports with “rice bodies” described by the radiologist. The initial search yielded a total of 100 patients between June 1992 and July 2022. Patients were excluded if their MRI was not available for retrospective review or if there was insufficient follow-up to establish a clear underlying diagnosis. Reports and corresponding MRI were reviewed by a musculoskeletal radiology fellow or a fellowship-trained musculoskeletal radiology attending with 13 years of subspeciality experience. The corresponding electronic medical records—including clinical notes, relevant laboratory values, and operative notes—were also reviewed to determine the underlying diagnosis attributed to rice body-like intraarticular particles on MRI. A total of 52 patients met the inclusion criteria and were enrolled in our retrospective study (Figure 3). Table 1 summarizes the demographic information of the study group. There were 32 females and 20 males within our cohort with a mean age of 52.8 (range 3-82) years.

Figure 3.

Flow-chart of study population selection.

Table 1.

Demographic Information for the 52 Patients With Rice Body-Like Synovial Particles on Magnetic Resonance Imaging. ^a

Age	52.8 (SD 22.4) y
Gender
Male	20 (38.5%)
Female	32 (61.5%)
Prior surgery
Arthroplasty	14 (26.9%)
Hip	11 (21.2%)
Knee	3 (5.8%)
Other surgery	5 (9.6%)
Rotator cuff repair	1 (1.9%)
Hamstring tendon repair followed by debridement	1 (1.9%)
Right hip incision and drainage	1 (1.9%)
Radial head excision. Interposition arthroplasty with Achilles tendon graft and ulnar nerve transposition	1 (1.9%)
Fifth metatarsal head resection	1 (1.9%)
None	33 (63.4%)
Location
Upper extremity	22 (42.3%)
Shoulder (glenohumeral) joint	9 (17.3%)
Elbow joint	5 (9.6%)
Wrist (radiocarpal) joint	3 (5.8%)
Subacromial subdeltoid bursa	3 (5.8%)
Olecranon bursa	1 (1.9%)
Second and third digit flexor tendon sheaths	1 (1.9%)
Lower extremity	29 (55.8%)
Hip	14 (26.9%)
Knee	10 (19.2%)
Forefoot bursa	2 (3.8%)
Anterior tibialis tendon sheath	1 (1.9%)
Extensor digiti minimi tendon sheath	1 (1.9%)
Ischial bursa	1 (1.9%)
Both upper and lower extremity (knee, shoulder, and wrist in 1 patient)	1 (1.9%)
Location
Joint	42 (80.8%)
Bursa	7 (13.5%)
Tendon sheath	3 (5.8%)

Abbreviation: SD, standard deviation.

^aValues are numbers of patients (%) or a mean and SD.

Standard MRI of the relevant joint was performed in either 1.5T or 3.0T scanners (GE Healthcare) utilizing routine pulse sequences according to our institution, typically consisting of axial, coronal, and sagittal intermediate-weighted fast spin echo sequences as well as a single inversion recovery sequence. However, in some cases, additional sequences were obtained at the time of exam as deemed appropriate by the supervising radiologist. Some patients in the study had previously undergone a total hip or knee arthroplasty; in these cases, additional multiacquisition variable-resonance image combination (MAVRIC) intermediate-weighted and inversion recovery sequences were acquired for the purposes of metal artifact reduction.

The MRIs were retrospectively reviewed by a musculoskeletal fellowship-trained radiology attending physician with 13 years of experience, who either agreed or disagreed with the reported presence of rice body-like synovial particles on MRI. These were defined as ovoid monomorphic bodies that were hypointense on intermediate-weighted sequences and located throughout or in dependent areas of the joint space, not confined to the peripheral synovial lining (Figure 4). Cases were excluded when the radiologist did not agree with the presence of rice body-like particles as described on the MRI report. A musculoskeletal pathologist with over 35 years of experience reviewed microscope slides from the surgical specimens available from our study cohort to determine the presence or absence of rice bodies, defined as papillary synovial hyperplasia with cores of fibrin or necrosis with or without inflammatory cells. Similar to previous reports, most rice bodies were in the range of 4 to 7 mm in greatest dimension. ²⁰ While there is histologic overlap between papillary hyperplasia of inflamed synovium and rice bodies, cases containing inflamed synovial villi <4 mm with no necrosis or fibrin were interpreted as synovial hyperplasia, not rice bodies. If no rice bodies were identified, the pathologist suggested other diagnoses based on the histological appearance (eg, synovial chondromatosis or chronic synovitis not-otherwise-specified).

Figure 4.

Rice body-like synovial particles in various joints and bursae on MRI. Coronal intermediate-weighted images of the (A) right shoulder, (B) left hip, and (C) left wrist in different patients demonstrate multiple monomorphic hypointense bodies in the subacromial-subdeltoid bursa, prosthetic hip joint and greater trochanteric bursa, and radiocarpal/midcarpal joints, respectively (white arrows).

MRI: magnetic resonance imaging.

Results

Of the 52 patients in our study, 37% (19/52) had prior surgery of the involved body part, most commonly (73%; 14/19) joint arthroplasty. There was upper extremity involvement in 42% (22/52) of patients, lower extremity involvement in 56% (29/52), and 1 patient had both upper and lower extremity involvement. No rice body-like particles were reported on MRI in any axial joints. Most rice body-like particles on MRI were seen in the joint proper (81%; 42/52) while a minority were seen in either a bursa (13%; 7/52) or tendon sheath (6%; 3/52). Rice body-like particles seen within bursae or tendon sheaths that were continuous with a joint space were considered to be located within the joint (eg, rice bodies located within the biceps tendon sheath of the shoulder).

Based on the clinical history, laboratory values, operative notes, and histopathological findings for the patients who subsequently underwent surgery after the MRI, rice body-like synovial particles were most commonly associated with an underlying diagnosis of inflammatory arthritis (52%; 27/52), followed by infection (25%; 13/52), synovial chondromatosis (8%; 4/52), osteoarthritis (6%; 3/52), bursitis (6%; 3/52), and adverse local tissue reaction (4%; 2/52) (Table 2). Among the 27 cases of inflammatory arthritis, 56% (15/27) were rheumatoid arthritis. Association with osteoarthritis and bursitis was made as a diagnosis of exclusion when no other diagnosis was identified.

Table 2.

Diagnoses Associated With Rice Body-Like Synovial Particles Seen on MRI (N = 52).

Inflammatory arthritis	27 (51.9%)
Rheumatoid arthritis	15
Juvenile idiopathic arthritis	7
Psoriatic arthritis	3
Overlap syndrome	1
Inflammatory arthropathy NOS	1
Infection	13 (25.0%)
Culture negative	4
Staphylococcus aureus	1
Other Staphylococcus species	5
Mycobacterium chelonae	1
Klebsiella	1
Corynebacterium	1
Synovial chondromatosis	4 (7.7%)
Osteoarthritis	3 (5.8%)
Bursitis	3 (5.8%)
Adverse local tissue reaction	2 (3.8%)

Abbreviations: MRI, magnetic resonance imaging; NOS, not otherwise specified.

Among the 13 infections, only 1 was mycobacterial, while 46% (6/13) were Staphylococcal and 31% (4/13) were culture-negative. The diagnosis of culture-negative infection was based on consultation with an infectious disease specialist. Twelve out of the 13 cases in which rice body-like particles on MRI were attributed to infection that occurred in patients with prior surgery in the affected joint. Among the 14 patients with joint arthroplasties (11 total hip arthroplasties and 3 total knee arthroplasties), rice body-like particles on MRI were associated with infection in 71% (10/14), followed by adverse local tissue reaction (14%; 2/14), and bursitis and inflammatory arthritis (7% [1/14] each) (Table 3).

Table 3.

Diagnoses Associated With Rice Body-Like Synovial Particles Seen on MRI in Patients Who Have Undergone Prior Arthroplasty Placement of the Joint (N = 14).

Diagnosis	Number (%) among those with arthroplasty
Prosthetic joint infection	10 (71%)
Adverse local tissue reaction	2 (14%)
Bursitis	1 (7%)
Inflammatory arthropathy	1 (7%)

Abbreviation: MRI, magnetic resonance imaging.

Among the study cohort, 46% (24/52) underwent surgery after MRI, and 92% (22/24) of those had slides still available for retrospective review by a pathologist for the purposes of this study. Out of the 22 patients with slides available for review, 10/22 (45%) had a histologic appearance consistent with rice bodies, in which 7 were associated with infection, 2 with inflammatory arthritis, and 1 with adverse local tissue reaction. Of the patients who were not found histologically to have rice bodies, 5/22 (23%) had papillary synovial hyperplasia with acute inflammation consistent with infection, 2/22 (9%) had chronic synovitis not-otherwise-specified, 4/22 (18%) had synovial chondromatosis, and 1 patient (1.9%) subsequently underwent an arthroplasty with findings of osteoarthritis without rice bodies. Histologically, there is some overlap between papillary synovial hyperplasia and rice bodies, but in our study, not all cases of rice body-like particles seen on MRI contained a core of necrosis or fibrin to fit our histologic criteria for rice bodies. Among the 2 patients without available surgical specimens, 1 patient had arthroscopic and surgical operative findings of “loose bodies,” although the pathology report did not specifically mention rice bodies.

Discussion

Although the description of rice bodies is usually made histologically or on gross inspection, ovoid monomorphic particulate debris that resemble rice can also be seen on MRI; radiologists may also refer to these as “rice bodies” in an MRI report. Because most previous descriptions of rice bodies and their appearance on MRI have been based on case reports, neither the true specificity of the MRI appearance for histological rice bodies nor the relative likelihood of the potential diagnoses when these are identified on MRI is known.^2,3,15,21 Based on the 22 cases in which surgery was performed after the MRI and histological slides were available for retrospective review, what looked like rice bodies on MRI were confirmed to be rice bodies on histology in 45% (10/22). Therefore, we suggest using the term “rice body-like particles” rather than “rice bodies” on MRI reports, as what looks like rice bodies on MRI do not always correspond to rice bodies on histology.

At our institution, a dedicated orthopedic hospital, when rice body-like particles are seen on an MRI, about half (52%) were attributed to an inflammatory arthritis, most commonly rheumatoid arthritis, and another 25% were attributed to infection, similar to prior reports.^1-3,10 Interestingly, although many of the published case reports of infection presenting as rice bodies represented mycobacterial infections, only 1 of the 13 cases of infection in our cohort was due to mycobacteria. Staphylococcus was the most common isolated organism in our study cohort, seen in 46% (6/13), while an organism could not be identified in 31% (4/13) of cases. We suspect that the incidence of mycobacterial infection associated with rice bodies depends on the frequency of mycobacterial infection in the general population, and a higher incidence is likely in regions where mycobacterial infections are endemic.

Synovial chondromatosis was the third most common diagnosis when rice body-like particles were seen on MRI. Histologically, synovial chondromatosis is composed of numerous more-or-less spherical nodules of hyaline cartilage, either embedded in synovium or free in the joint space. These nodules have a size range that overlaps with that of rice bodies and may have an MRI appearance indistinguishable from rice bodies when they are composed of pure cartilage, but they are more readily differentiated from rice bodies when the cartilage nodules become partially mineralized. ³ In the absence of mineralization, distinguishing between synovial chondromatosis and rice bodies is difficult with MRI alone. One prior study has suggested that T1-weighted imaging can distinguish between these 2 entities, as rice bodies are isointense to surrounding fluid on T1-weighted imaging while synovial chondromatosis appears slightly T1-hypertintense. ³ At our institution, we utilize intermediate-weighted sequences to optimize cartilage signal and T1-weighted sequences are not routinely included for MRI of the extremities; thus, this phenomenon could not be evaluated on the current study.

Within our cohort, osteoarthritis and bursitis were other, less common, diagnoses associated with rice body-like particles on MRI, although these were diagnoses of exclusion. Several case reports have described the presence of florid rice bodies in patients without underlying infection or rheumatologic disease^14,15 and others have described rice bodies in cases of tenosynovitis ²² and chronic bursitis.^3,17,23,24 While osteoarthritis has not classically been associated with rice body formation, there are case reports describing this phenomenon ²⁵ ; it is also possible that a rare subset of osteoarthritis can secrete inflammatory mediators that may progress more rapidly and overlap with other inflammatory arthropathies. ²⁶

In the presence of an arthroplasty, case reports have described rice bodies in association with adverse local tissue reaction, ⁴ infection, ²⁷ and in large synovial cysts without infection.^19,20 Our study shows that the presence of rice body-like particles on MRI in the setting of a joint arthroplasty should be considered prosthetic joint infection until proven otherwise, as the incidence of infection was 71% in our arthroplasty cohort.

Our study had several limitations. First, to identify our study cohort we relied on the keyword “rice” in the MRI reports; however, not all radiologists (even at our own institution) use “rice body” as a descriptor. Thus, there may be more cases of rice body-like particles on MRI than our search identified. Second, our gold standard to compare the MRI cases was histology rather than gross inspection, as gross specimens were not available for retrospective review. There is no universal definition of what constitutes a rice body, but historically the description has been based on gross inspection rather than histology. Thus, using the histologic criteria (papillary synovial villi with a core of fibrin or necrosis) for the definition may have resulted in underestimation of the true incidence of rice bodies among the subgroup who underwent surgery. Finally, the frequency of certain diagnoses is dependent on the patient population, thus our results may not apply to a population with a lower percentage of postarthroplasty MRI or in a region where mycobacterial infection is endemic.

In conclusion, rice body is a histopathologic description of fibrinous loose bodies historically associated with tuberculosis and inflammatory arthritis, especially rheumatoid arthritis. When rice body-like synovial particles are identified on MRI, a large minority corresponded to rice bodies on histology. In our study cohort, inflammatory arthritis and nonmycobacterial infection were the most common diagnoses associated with rice body-like particles on MRI, but in patients with an arthroplasty, nonmycobacterial infection was by far the most common diagnosis, followed by adverse local tissue reaction.