Villonodular synovitis (PVNS) of the spine
- First Online:
- Cite this article as:
- Motamedi, K., Murphey, M.D., Fetsch, J.F. et al. Skeletal Radiol (2005) 34: 185. doi:10.1007/s00256-004-0880-9
To describe the imaging features of spinal pigmented villonodular synovitis (PVNS).
Design and patients
We retrospectively reviewed 15 cases of pathologically proven spinal PVNS. Patient demographics and clinical presentation were reviewed. Radiologic studies were evaluated by consensus of two musculoskeletal radiologists for spinal location, spinal segments affected, lesion center, detection of facet origin and intrinsic characteristics on radiography (n =11), myelography (n =7), CT (n =6) and MR imaging (n =6).
Women (64%) were more commonly affected than men (36%) with an average age of 28 years. Clinical symptoms were pain (45%), neurologic (9%) or both (36%). Lesions most frequently affected the cervical spine (53%) followed by the thoracic (27%) and lumbar regions (20%). The majority of lesions (93%) were centered in the posterior elements with frequent involvement of the pedicle (67%), neural foramina (73%), lamina (67%) and facets (93%). No lesions showed calcification. Determination of a facet origin by imaging was dependent on imaging modality and lesion size. A facet origin could be determined in 45% of cases by radiography vs 67% of patients by CT (n=6) and MR (n=6). Large lesions (greater than 3 cm in at least one dimension) obscured the facet origin in all cases with CT and/or MR imaging (44%,n=4). Small lesions (less than 3 cm in any dimension) demonstrated an obvious facet origin in all cases by CT and/or MR imaging (56%,n=5). Low-to-intermediate signal intensity was seen in all cases on T2-weighted MR images resulting from hemosiderin deposition with “blooming effect” in one case with gradient echo MR images.
PVNS of the spine is rare. Large lesions obscure the facet origin and simulate an aggressive intraosseous neoplasm. Patient age, a solitary noncystic lesion centered in the posterior elements, lack of mineralization and low-to-intermediate signal intensity on all MR pulse sequences may suggest the diagnosis in these cases. Small lesions demonstrate a facet origin on CT or MR imaging. This limits differential considerations to synovial-based lesions and additional features of a solitary focus, lack of underlying disease or systemic arthropathy, no calcification as well as low-to-intermediate signal intensity on all MR images should allow spinal PVNS to be suggested as the likely diagnosis.
Pigmented villonodular synovitis (PVNS) represents a slowly progressive tumefactive proliferative process of the synovium of unknown etiology [1–4]. There are two basic types of PVNS, one affecting the synovium of a joint diffusely, while the localized form typically involves the synovium focally about a tendon sheath [1–4]. The diffuse form of PVNS is usually monoarticular and most frequently affects the knee, followed by the hip, ankle, shoulder and elbow [1–4]. The localized or focal form of PVNS, also referred to as giant cell tumor of the tendon sheath (GCTTS), most commonly involves the hand, foot and knee [5–10]. PVNS has an estimated incidence of two per million persons annually [1–4, 11].
The spine is a very unusual site of origin for PVNS, and thus it is a diagnosis that is frequently not considered at this anatomic location. These lesions typically arise from the synovium of the facet joints. There is only limited scientific literature (most isolated case reports) evaluating spinal PVNS [12–28]. Giannini and coworkers reviewing 12 cases and Furlong and colleagues reporting 15 cases represent the largest series of PVNS involving the spine to date [19, 28]. However, both of these studies emphasized the pathologic appearance of spinal PVNS, with only limited discussion and description of the imaging manifestations and spectrum [19, 28]. These studies were also largely limited to radiographic evaluation, although computed tomography (CT) and magnetic resonance (MR) images were illustrated.
The purpose of our study is to comprehensively review the imaging characteristics of 15 pathologically proven cases of spinal PVNS.
Materials and methods
We retrospectively reviewed 15 cases of pathologically proven spinal PVNS from our institution (Armed Forces Institute of Pathology). This study was performed with approval of the (Armed Forces Institute of Pathology) Human Subjects Committee. Informed consent was not required. The patient demographics including gender and age as well as the presenting clinical symptoms were recorded. The radiologic material available for review included radiographs (n=11), conventional tomography (n=4), myelography (n=7), angiography (n=1), CT (n=6), and MR images (n=6). Two musculoskeletal radiologists (KM, MDM) reviewed the imaging material with agreement by consensus.
The imaging material was evaluated for the general spinal area involved (cervical, thoracic or lumbar) and the specific level affected. Lesion size was determined by the best imaging modality available. At each level the spinal segments affected (any segment involved by the lesion-vertebral body, pedicle, neural foramina, lamina, facet, spinous process, transverse process, costovertebral joint and paraspinal soft tissues) was determined using the best imaging modality depicting the lesion. Lesion center (vertebral body, posterior elements or paraspinal soft tissues) was determined using the best imaging modality depicting the lesion. An additional evaluation if a facet origin of the lesion could be identified (lesion centered in the facet with erosions of both sides of the joint) or not was also performed with radiography, CT and MR imaging. In cases with CT or MR imaging, the ability to identify a facet origin of the lesion was compared to lesion size (less than or greater than 3 cm in at least one dimension).
Radiographs and conventional tomograms were evaluated together for type of bone lysis if present (geographic or moth-eaten/permeative) and zone of transition (narrow or wide), presence and extent of sclerotic rim (and, if present, the extent graded as less than 25%, 25–50%, 50–75% or more than 75%), presence or absence of calcification, internal trabeculation and expansile remodeling of bone (and degree, if present, graded as mild, moderate or marked). The myelograms were evaluated for presence or absence of an epidural soft-tissue mass and/or total myelographic block. Angiography was evaluated for degree of vascular staining, identification of the supplying arteries and presence of early draining veins.
CT was evaluated for lesion margin as defined with a capsule (complete definable high attenuation rim), defined without a capsule or partial capsule (no or incomplete definable high attenuation rim) and infiltrative (irregular and ill-defined). The predominant attenuation was determined compared with paravertebral muscles (lower than muscle, similar to muscle or higher than muscle) and whether the lesion was homogeneous or heterogeneous (if heterogeneous, the degree was graded as mild, moderate or marked). The presence or absence of an associated soft-tissue mass was evaluated. For lesions affecting the cervical spine, involvement of the foramen transversarium and the vertebral artery was determined. The lesions were also evaluated for the presence or absence of trabeculation (if present, the degree was graded as mild, moderate or marked), calcification, sclerotic margin (if present, the extent was graded as surrounding less than 25%, 25–50%, 50–75% or more than 75% of the lesion), and osseous expansile remodeling of bone (if present, the degree was graded as mild, moderate or marked).
The MR images available were all performed with high-magnetic-field MR units (1.0 T or 1.5 T), and included at least one short TR/TE (range 575–800/12–20 ms) (n=6) pulse sequence and either a long TR/TE (1,500–4,619/98–112 ms) (n=5) pulse sequence or a T2*weighted (TR 25/TE 15, flip angle 20°) (n=1) pulse sequence. Lesion margins were assigned to one of three categories, either defined with a capsule (complete definable low-intensity rim on all pulse sequences), defined without a capsule or with a partial capsule (no or incomplete definable low-intensity rim on all pulse sequences) and infiltrative (irregular and ill-defined). The predominant T1-weighted signal intensity (low: lower than muscle; intermediate: similar to muscle; high: similar to fat) and T2-weighted signal intensity (low: similar to muscle; intermediate: similar to fat; high: greater than fat) of the lesion was also evaluated. In addition, the percentage (5–25%, 25–50%, 50–75%, 75–100%) of lesion that was low, intermediate or high signal on T2-weighted MR images was determined (to be considered a significant component, at least 5% of the lesion had to be composed of that signal-intensity tissue). Both sequences were also reviewed for lesion homogeneity or heterogeneity (if heterogeneous, the degree was graded as mild, moderate or marked). The single case with T2* images was also evaluated for the presence of “blooming” and its degree (graded as mild, moderate or marked). The presence or absence of associated surrounding bone marrow and/or soft-tissue edema was evaluated (on long TR/TE or T2* sequences only) and, if present, the degree was graded as mild, moderate or marked. The presence or absence of an accompanying soft-tissue mass was also noted. For lesions affecting the cervical spine, involvement of the foramen transversarium and vertebral artery was evaluated. The lesions were also evaluated for the presence of fluid levels and/or areas of high signal on both short and long TR/TE sequences representing hemorrhagic foci. The degree (graded as mild, moderate, marked) and pattern (diffuse, peripheral, peripheral/septal, peripheral/nodular, nodular) of enhancement in cases with gadolinium ( n=2) were also evaluated.
PVNS Spine: demographics, symptoms and lesion size/location ( VB vertebral body, P pedicle, NF neural foramina, L lamina, F facet, SP spinous process, ST soft tissue, CND could not be determined as no measuring parameters were available on archived images)
C3–4 / P, NF, L, F, ST
L4–5 / L, F, ST
C5–6 / VB, P, NF, L, F, ST
T4–5 / P, NF, L, F, ST
C3–4 / P, NF, L, F, ST
C2–3 / P, NF, L, F, ST
C4–5 / P, NF, F, ST
C5–6 / VB, NF, L, F, SP, ST
C5–6 / VB, P, NF, L, F, ST
L5-S1 / F, ST
T2–3 / P, NF,F, ST
T5–6 / P, NF, F, ST
T 5–6 / ST(paraspinal)
L4–5 / L, F, ST
C4–5 / VB, P,NF, L,F, ST
Imaging findings: PVNS spine ( TZ transition zone, Ho homogenous, He heterogeneous, NA not applicable)
Wide TZ 9 (n =90%)
Defined 5 (83%)
Defined 6 (100%)
Ill-defined 1 (17%)
T1-intermediate 6 (100%)
T2-intermediate 4 (67%), low 2 (33%)
Similar to muscle 6 (100%)
Ho 4 (67%)
T1-Ho 1(17%); He 5 (83%)
He 2 (33%)
T2 - He 6 (100%)
Facet origin identified
CT revealed a defined margin without a capsule in 83% (n=5) of cases (Fig. 3A, 6C) and in one case (17%) showed an infiltrative margin. In all cases the predominant attenuation of the lesion was similar to the paraspinal muscles (Figs. 3A, 6C). Lesions were homogenous in 67% (n=4) of cases and heterogeneous in 33% (n=2, one mild and one moderate) of cases. CT showed an associated soft-tissue mass in all cases (n=6). With CT, in 50% (n=2) of the cases affecting the cervical spine (n=4), the lesion extended into the foramen transversarium to involve the vertebral artery (Fig. 1D). No cases demonstrated internal trabeculation or calcification (Figs. 1D, 6C). A sclerotic margin (surrounding less than 25% of the lesion) was seen in only one case (17%). Mild-to-moderate expansile remodeling of the osseous structures was observed in 67% (n=4) of the cases (Figs. 1D, 6C).
MR imaging revealed a defined margin without a capsule or only a partial capsule in all cases (100%,n=6) (Fig. 4). On T1-weighted MR images all lesions were of intermediate (similar to muscle) signal intensity (Figs. 2A, 3B). One case (17%) was homogenous; four (67%) were mildly heterogeneous, and one (17%) was moderately heterogeneous. On T2-weighted or T2* weighted images, the predominant signal intensity was intermediate (similar to fat) in 67% (n=4) of cases and low (n=2, similar to muscle) in 33% of cases (Figs. 2C and 4). Significant areas (>5% of the lesion) of low signal intensity on T2-weighted MR images were seen in three of the four cases (two cases 5–25% and one case 25–50%) with predominant intermediate-signal-intensity tissue. Significant areas of high-signal-intensity tissue were seen in two cases (one case 5–25%; one case 25–50%) on T2-weighted MR images. Lesions showed heterogeneous signal intensity in all cases on T2 weighting (mild in four and moderate in two). The single case with T2*weighted images demonstrated a mild blooming effect. No cases revealed surrounding bone-marrow edema. However, 40% (n =2) of cases showed edema of the surrounding soft tissues (one marked, one mild). In one case the surrounding soft tissues were artifactually too high in signal intensity to evaluate for surrounding edema. MR imaging showed a soft-tissue component in 83% (n=5) of cases (Fig. 3B). With MR imaging, in 67% (n=2) of the cases affecting the cervical spine (n=3) the lesion extended into the foramen transversarium to involve the vertebral artery (Fig. 2). None of the lesions contained fluid levels or foci of high signal on both short and long TR/TE sequences to suggest the presence of hemorrhage. In the two cases imaged following administration of intravenous gadolinium, there was diffuse enhancement of the lesions (moderate in both) (Fig. 2B).
In 1941 Jaffe and colleagues coined the term PVNS to describe a proliferative, tumefactive villonodular lesion of the synovium . Since this original description there have been numerous publications describing the imaging characteristics of PVNS in the appendicular skeleton [4, 5, 11]. However, PVNS of the axial skeleton is only rarely reported [12–28]. With the exception of Giannini and colleagues in 1996 reviewing 12 cases and Furlong and coworkers’ report of 15 cases, only sporadic cases of this entity have been reported in the literature [19, 28]. These two largest reports emphasized the pathologic appearance of spinal PVNS, with only limited discussion and description of the imaging manifestations [19, 28]. In addition, we believe that several of the isolated case reports, particularly in the lumbar spine, may have mistakenly interpreted the pathologic appearance of degenerative facet disease with hemorrhage and detritic debris as PVNS [14, 15, 26, 29]. Indeed, hypertrophic synovitis associated with facet degenerative disease is common and has been reported to cause paraarticular masses in 2% of lumbar laminotomy procedures by Savitz and colleagues [25, 30–33]. In our opinion, the case reported as PVNS at L4–5 by Titelbaum and colleagues is a typical example, with the CT demonstrating prominent bilateral facet degenerative changes (without destruction of the facet as expected for PVNS) and the medial unilateral epidural mass revealing calcification caused by hypertrophic synovitis and cyst formation filled with osteochondral debris and reactive tissue. The distinction of these entities, pathologically, is that hypertrophic synovitis reveals only hyperplasia of the synovium (synoviocytes confined to villous surfaces), as opposed to the proliferation of synovial cells and abundant giant cells in PVNS [25, 28, 31, 32, 34, 35]. The fact that these pathologic differences may be subtle serves to emphasize the need for radiologic-pathologic correlation in diagnosis of musculoskeletal lesions .
The analysis of our clinical data reveals both similarities and discrepancies with previous reports. Our study showed that spinal PVNS was more common in females (64%) than males. This is in contrast to reports of equal sex distribution when considering PVNS at other sites [1–4]. Our data support the wide range of age distribution with an average age in the third or fourth decade, as previously described [1–4]. Previous reports indicate a predilection for the cervical or lumbar spine [12–28]. Our study also confirms the predilection for the cervical spine (53% of cases). However, unlike previous reports, the lumbar spine was the least common spinal region affected, with only 20% of lesions at this site, while 27% of our cases involved the thoracic spine [12–28]. This may be due to the fact that the cervical lesions may be more symptomatic and thus come to attention more readily than thoracolumbar spine lesions, as the thoracic spine otherwise numerically contains the largest absolute numbers of facet joints. An additional importance to the cervical spine predilection is the significant incidence of involvement of the foramen transversarium and vertebral artery (50% of cases by CT and 67% by MR), with obvious surgical implications for the ability to perform complete resection.
Not surprisingly, radiography less commonly (45%) identifies the facet joint as the site of lesion origin, compared with CT (67%) or MR imaging (67%). This is related to the complex anatomy of the spinal column and failure to profile the facet joints adequately. Oblique radiographs are optimal for visualizing the facet joints in the cervical and lumbar regions, although currently these projections are often not routinely obtained. However, the lesion center in the posterior elements was demonstrated on imaging in 93% of cases.
CT and MR imaging revealed well-defined lesions in 83% and 100% of lesions, respectively. Lesions were more commonly homogenous on CT (67% of cases), as compared with heterogeneity on MR imaging (84% on T1 weighting, 100% on T2 weighting). The predominant intrinsic appearance of spinal PVNS was soft-tissue attenuation on CT (100%) and intermediate signal intensity on T1-weighted MR (100%) and low (33%) to intermediate (67%) signal intensity on T2-weighted MR. In addition, 83% of lesions showed significant foci (at least up to 25% of the lesion volume) of low signal intensity on T2-weighted MR images, reflecting hemosiderin deposition. Associated soft-tissue masses were identified in all cases (n=6) by CT and in 83% of cases (n =5) by MR imaging. No areas of calcification or hemorrhage were apparent on CT or MR imaging. Moderate diffuse enhancement was seen in the lesion in both cases with MR imaging, following contrast administration caused by increased vascularity, as has been depicted in appendicular PVNS [6–10, 12–19, 21, 22, 24–35].
CT and MR imaging revealed two basic patterns of radiologic presentation primarily based on lesion size. Large lesions (greater than 3.0 cm in at least one dimension) represented 44% (n=4) of our cases with either CT or MR imaging for review. These large lesions demonstrated an aggressive appearance involving many osseous spinal segments. Although all imaging modalities are able to localize the pathologic process to the posterior elements, recognition that these large lesions originated in the facet joint is difficult to impossible, in our opinion and experience, even with the use of cross-sectional imaging, because of this extensive involvement.
The differential diagnosis for these large spinal PVNS lesions primarily includes osseous neoplasms that originate in the posterior elements, such as osteoblastoma, aneurysmal bone cyst (ABC), metastatic disease, multiple myeloma, lymphoma, giant cell tumor (GCT), chordoma and chondrosarcoma . This is again a reflection of the large lesion size and multiple osseous spinal segments (facet, lamina, spinous process, pedicle, and vertebral body) affected. This extensive differential diagnosis can be more restricted by analysis of other intrinsic imaging features. Lack of mineralized matrix or calcification (best evaluated by CT) limits the possibility of osteoblastoma, chordoma and chondrosarcoma [37–45]. An ABC invariably contains prominent or entirely cystic components, with multiple fluid levels revealing high signal intensity on T2-weighted MR images and marked aneurysmal expansile remodeling of bone [46–49]. The younger age of the patient and the solitary focus of the lesion make the diagnoses of metastatic disease, multiple myeloma and lymphoma unlikely. In addition, these lesions are much more commonly centered in the vertebral body. GCT, while often showing similar signal intensity to spinal PVNS, is typically centered in the vertebral body, as opposed to the posterior elements [50, 51]. Chordoma (centered in the vertebral body, with a large associated soft-tissue mass) and chondrosarcoma (centered in vertebral body or posterior elements and large associated soft-tissue mass) display high signal intensity on T2-weighted MR images because of their intrinsically high water content [40–45]. Thus, predominant low-to-intermediate signal intensity on all MR pulse sequences in a large, solitary posterior element lesion affecting the spine and lacking calcification in a young patient may be the only imaging feature to suggest the diagnosis of PVNS originating in the apophyseal joints. We had only one case in our series with gradient echo MR images (T2*) available for review. It showed a mild “blooming” effect, as the result of hemosiderin, and this characteristic may also be helpful for diagnosis. CT and MR imaging are also useful to detect involvement of the foramen transversarium and vertebral artery (seen in 50% of our cervical spine cases by CT and 67% by MR imaging), an important feature in planning surgical resection.
Small lesions (smaller than 3.0 cm in all dimensions) represented 56% (n=5) of our cases with either CT or MR imaging available for review. In contradistinction to large spinal PVNS lesions, CT and MR imaging of smaller lesions identify the facet origin of the pathologic process, because involvement is relatively localized to the apophyseal joint causing erosion on both sides of the articulation. This is vital information to recognize because of its implications for the differential diagnosis, limiting considerations to synovial-based processes, just as in the appendicular skeleton. This differential diagnosis includes lesions such as gout, calcium pyrophosphate deposition (CPPD) disease, amyloid deposition (secondarily related to chronic renal failure), inflammatory arthropathy (rheumatoid arthritis and the seronegative arthropathies), synovial chondromatosis and PVNS. Gout and the inflammatory arthritides affect the spine quite late in the disease process. Invariably these demonstrate systemic polyarticular manifestations, with longstanding disease by the time spinal involvement is apparent [52–54]. CPPD arthropathy and synovial chondromatosis usually reveal calcifications best demonstrated by CT [53, 55]. CPPD rarely affects the spine, and chondrocalcinosis should be seen in the typical sites of the knee and wrist [53, 55]. Synovial chondromatosis is exceedingly rare in the spine and reveals high signal intensity on T2-weighted MR images, owing to the high water content of non-mineralized hyaline cartilage [11, 56, 57]. Amyloid arthropathy related to chronic renal failure commonly affects the spine (destructive spondyloarthropathy) but is usually centered in the disk, as opposed to the facet joints as in PVNS, and it is often multifocal [58, 59]. All of these synovial diseases (except synovial chondromatosis) may show predominant intermediate-to-low signal on all MR pulse sequences related to chronicity and fibrosis similar to the intrinsic characteristics of spinal PVNS.
We acknowledge the limitations of our study. A relatively small number of patients were imaged, particularly with CT and MR, although this remains the largest reported group of spinal PVNS evaluated radiologically. The imaging techniques, modalities, and parameters were not standardized and varied significantly, owing to the multiple centers referring cases, and the specific neurologic symptoms present (spinal-cord related or nerve root) were often not stated. Post-contrast imaging and gradient echo MR sequences were available in only a minority of patients. Pathologic material was often limited and could not be mapped to directly compare to imaging. However, imaging and pathologic material that was available was considered adequate for evaluation. Despite these limitations, we believe our findings add further information to the understanding of PVNS affecting the spine.
In conclusion, although PVNS only rarely affects the spine, radiologic features may suggest the diagnosis. Large lesions (greater than 3 cm in any dimension) obscure the facet origin and simulate an aggressive intraosseous neoplasm. Patient age, a solitary noncystic lesion centered in the posterior elements, lack of mineralization and low-to-intermediate signal intensity on all MR pulse sequences may suggest the diagnosis. In small lesions (less than 3 cm in all dimensions) a facet origin can be identified on CT or MR imaging, as the lesion is often confined to this joint. This limits differential considerations to synovial-based lesions. Additional features of a solitary focus, lack of underlying disease or systemic arthropathy, no calcification as well as low-to-intermediate signal intensity on all MR images should allow spinal PVNS to be suggested as the likely diagnosis.
The authors gratefully acknowledge the support of Anika Ismel Torruella for manuscript preparation. In addition, we thank the residents, without whom this project would not have been possible, who attend the Armed Forces Institute of Pathology’s radiologic-pathology courses (past, present, and future) for their contribution to our series of patients.