Inflammation and angiogenesis in osteoarthritis




To quantify the relationship between inflammation and angiogenesis in synovial tissue from patients with osteoarthritis (OA).


Hematoxylin and eosin staining and histologic grading for inflammation were performed for 104 patients who met the American College of Rheumatology criteria for OA and had undergone total joint replacement or arthroscopy. A purposive sample of synovial specimens obtained from 70 patients was used for further analysis. Vascular endothelium, endothelial cell (EC) proliferating nuclei, macrophages, and vascular endothelial growth factor (VEGF) were detected by immunohistochemical analysis. Angiogenesis (EC proliferation, EC fractional area), macrophage fractional area, and VEGF immunoreactivity were measured using computer-assisted image analysis. Double immunofluorescence histochemical analysis was used to determine the cellular localization of VEGF. Radiographic scores for joint space narrowing and osteophyte formation in the knee were also assessed.


Synovial tissue samples from 32 (31%) of 104 patients with OA showed severe inflammation; thickened intimal lining and associated lymphoid aggregates were often observed. The EC fractional area, EC proliferation, and VEGF immunoreactivity all increased with increasing histologic inflammation grade and increasing macrophage fractional area. In the synovial intimal lining, VEGF immunoreactivity was localized to macrophages and increased with increasing EC fractional area and angiogenesis. No inflammation or angiogenic indices were significantly correlated with radiographic scores.


Inflammation and angiogenesis in the synovium are associated with OA. The angiogenic growth factor VEGF generated by the inflamed synovium may promote angiogenesis, thereby contributing to inflammation in OA.

Osteoarthritis (OA) is the commonest joint disease in the world, and up to 50% of the US population older than age 50 years may be affected (1). Symptoms of OA include chronic pain and disability (1). An inflammation component of OA has been observed (2) but is often considered of secondary importance to joint damage. Elevations in serum acute-phase proteins, such as C-reactive protein (3), have also been reported in patients with OA.

Chronic synovial inflammation in rheumatoid arthritis (RA) is associated with hyperplasia of the intimal lining and results in rapid tissue turnover, with high rates of endothelial cell (EC) proliferation and apoptosis (4, 5). The potential importance of synovial neovasculature in OA remains largely unrecognized (6), although some individuals may display indices of EC turnover as high as those observed in the synovium of patients with RA (7).

Vascular endothelial growth factor (VEGF) is a potent stimulator of angiogenesis and may also contribute to inflammation via plasma extravasation (8). Different splice variants may be derived from the VEGF gene. The major forms are believed to be VEGF121, VEGF165, and VEGF189, all of which share a common amino terminus (8). Synovial lining cells derived from patients with OA may release VEGF (9), raising the possibility that synovial tissue–derived VEGF may contribute to angiogenesis and inflammation in OA. We propose that subclinical synovitis is common in OA, and that it is associated with increased rates of angiogenesis in osteoarthritic knees and hips.



Synovial tissue samples obtained from 104 patients (51 women, 53 men, median age 69 years [interquartile range 62–75 years]) with OA of the knee (n = 71) or hip (n = 33) were collected at the time of total joint replacement surgery (n = 97) or arthroscopy (n = 7). All participants fulfilled the American College of Rheumatology revised criteria for the diagnosis of OA (10, 11), and informed consent was obtained prior to sample collection. The study protocol was approved by the North Nottinghamshire Health Authority Local Research Ethics Committee (project NNHA/420).

Synovial tissue samples from a purposive sample of 70 patients, representative of the full range of synovial inflammation (22 hip specimens and 48 knee specimens, including 4 arthroscopic samples), were then selected for quantitative histomorphometry. Synovial tissue samples from an additional purposive sample of 20 patients (10 with no synovial inflammation, 10 with severe inflammation) were selected for double fluorescence immunohistochemical analysis. Five (5%) of these 104 patients reported previous surgical trauma to the affected joint. None of the patients were receiving corticosteroids or slow-acting antirheumatic drugs.

Radiographic determination of disease severity.

Radiographs, obtained prior to surgery, were available for 40 of the 48 patients with OA of the knee. Radiographic scores for joint space narrowing (JSN) and osteophytes of the sampled knee were assigned to each case, as described by Nagaosa et al (12). Total JSN scores and total osteophyte scores were defined as the sum of the respective scores for each compartment of a joint (12).

Sample preparation and grading.

Synovium was coated with Tissue Tek (RA Lamb, Eastbourne, UK) and frozen in melting isopentane. Four-micrometer sections were used for hematoxylin and eosin (H&E) staining and were histologically graded according to the following scale: 0 = normal (synovial lining <4 cells thick, sparse cellular distribution, with few or no inflammatory cells), 1 = mild inflammation (synovial lining 4 or 5 cells thick, increased cellularity with some inflammatory cells), 2 = moderate inflammation (synovial lining 6 or 7 cells thick, dense cellularity with inflammatory cells but not lymphoid aggregates), 3 = severe inflammation (synovial lining >7 cells thick, dense cellularity and inflammatory cell infiltration, may contain perivascular lymphoid aggregates) (Figures 1A–D). Grades were assigned by 2 independent observers. When grades differed, the synovium was reexamined, and a consensus was reached.

Figure 1.

Inflammation and vascular endothelial growth factor (VEGF) immunoreactivity in synovial tissue obtained from patients with osteoarthritis (OA). A–D, Hematoxylin and eosin–stained sections showing each grade of inflammation. A, Grade 0 (apparently normal tissue). B, Grade 1 (mild inflammation). C, Grade 2 (moderate inflammation). D, Grade 3 (severe inflammation). E and F, Double immunofluorescence histochemical analysis for VEGF (E) and macrophages (F) in one section of severely inflamed synovium from a patient with OA. G and H, Double immunofluorescence histochemical analysis for VEGF (G; long arrows indicate nonfibroblast vessel cells with VEGF immunoreactivity) and fibroblasts (H) in a section of severely inflamed synovium from a different patient with OA (arrowheads indicate cells without colocalization of immunoreactivity). Short arrows in E-H indicate cells with colocalization of immunoreactivity. Open arrows in A-H indicate the synovial lining. Bar = 100 μm.


Macrophages were visualized using mouse anti-human CD14 (clone UCHM-1; Sigma, Poole, UK). VEGF was visualized using a rabbit polyclonal antibody directed to the 20–amino acid terminal residues of VEGF (A20; Santa Cruz Biotechnology, Santa Cruz, CA). Primary antibodies were visualized using an avidin–biotin–peroxidase complex (ABC Elite kit; Vector, Peterborough, UK) and developed with FAST diaminobenzidine (DAB; Sigma) (7).

Double sequential immunohistochemical analysis was performed as previously described (4). Briefly, proliferating cell nuclei were visualized using anti–Ki-67 (clone MIB-5; Dako, Cambridge, UK) and the ABC Elite kit and then developed using nickel-enhanced DAB (4). Endothelium was then visualized using mouse monoclonal anti-CD31 (clone TLD-3A12; Serotec, Oxford, UK) and an alkaline phosphatase complex and developed using fast red (Sigma).

The double immunofluorescence technique was used to determine colocalization of VEGF immunoreactivity with macrophages or fibroblasts. Tissue sections were incubated with antibodies to VEGF, then with biotinylated goat anti-rabbit IgG, as described above, and then with Texas Red–streptavidin (Vector). Incubation with the second primary antibody (mouse anti-human prolyl 4-hydroxylase for fibroblasts [clone 5B5; Dako] or anti-CD14) was then performed. Sections were incubated with horse anti-mouse fluorescein isothiocyanate conjugate (Vector). Double immunohistochemical analysis was followed by incubation with 4′,6-diamidino-2-phenylindole dihydrochloride (Sigma) (4).

Image analysis.

Image analysis was performed according to the method described by Walsh et al (7), using a KS300 image analysis system (Imaging Associates, Thame, UK). Ki-67–positive ECs were identified as only those immunoreactive nuclei surrounded entirely by CD31 immunoreactivity. The EC proliferation index was defined as the percentage of ECs with Ki-67 immunoreactivity. The EC, VEGF, and macrophage fractional areas were defined as the fraction of synovial area immunoreactive for that particular antigen.

Statistical analysis.

Data were analyzed using SPSS software, version 8 (Chicago, IL). A Kruskal-Wallis nonparametric analysis of variance was performed to compare the dependent variables (EC proliferation index and EC fractional area) between inflammation grades. Possible associations between EC proliferation index, EC fractional area, VEGF immunoreactive and macrophage fractional areas, and radiographic scores were analyzed by Spearman's rank correlation. Data are presented as the median and interquartile range (IQR).


Synovial inflammation grades.

Of the original H&E-stained sections examined, samples from 32 (31%) of the 104 patients with OA showed evidence of severe (grade 3) synovial inflammation. Seven synovial samples from patients with OA were assigned inflammation grade 0, 29 were assigned grade 1, and 36 were assigned grade 2. Many of the most severely inflamed synovial samples (grade 3) had histologic features similar to those observed in synovium obtained from patients with RA, including the presence of lymphoid aggregates. Inflammation grades 2 and 3, respectively, were observed in 2 of 7 arthroscopy samples.

Macrophage infiltration.

The CD14 immunoreactivity was predominantly localized to the synovial lining and was particularly evident in the samples that showed the most inflammation. The median macrophage fractional area for all synovial samples obtained from patients with OA was 9.1% (IQR 4.4–16.9%). The macrophage fractional area increased with increasing histologic inflammation grade (r = 0.42, P < 0.01) (Table 1).

Table 1. Associations between indices of angiogenesis, inflammation, and VEGF immunoreactivity in synovial samples from patients with osteoarthritis*
 All samples (n = 70)Hips (n = 22)Knees (n = 48)
EC prolif indexEC fract areaInflam gradeMacro fract areaEC prolif indexEC fract areaInflam gradeMacro fract areaEC prolif indexEC fract areaInflam gradeMacro fract area
  • *

    Values are the correlation coefficients. VEGF = vascular endothelial growth factor; EC prolif index = endothelial cell proliferation index; EC fract area = EC fractional area; Inflam grade = inflammation grade; Macro fract area = macrophage fractional area.

  • P < 0.01.

  • P < 0.05.

EC fractional area0.38   0.54   0.32   
Inflammation grade0.590.48  0.660.4  0.550.52  
Macrophage fractional area0.370.40.42 0.710.440.56 0.170.410.37 
VEGF immunoreactivity fractional area0.

Angiogenesis and vasculature.

The EC fractional area and proliferation indices were examined to determine whether there were significant differences across the inflammation grades. The median EC fractional area for all synovial samples from patients with OA was 0.7% (IQR 0.5–1.1%). The median EC proliferation index for all synovial samples from patients with OA was 2.2% (IQR 0.4–3.8%). Significant heterogeneity between inflammation grades was observed for the EC proliferation index and the EC fractional area (χ2 = 27.7 and χ2 = 17.6, respectively, P < 0.001). The highest proportion of Ki-67–immunoreactive EC nuclei was observed in synovia with severe inflammation, and these were located adjacent to and within the lining layer (Figure 2). The EC fractional area was positively correlated with inflammation grade and macrophage fractional area (r = 0.48, P < 0.01 and r = 0.40, P < 0.01, respectively) (Table 1). The EC proliferation index was also positively correlated with the inflammation grade and macrophage fractional area (r = 0.59, P < 0.01 and r = 0.37, P < 0.01, respectively) (Table 1).

Figure 2.

Double sequential immunohistochemical analysis for nuclear proliferation (Ki-67, black staining) and endothelium (CD31, red staining) in noninflamed (A) and inflamed (B) synovia obtained from patients with osteoarthritis. Non–endothelial cell (EC) proliferating nucleus (arrowhead) is seen in noninflamed synovium (A), and proliferating EC nuclei (arrows) surrounded by EC immunoreactivity are seen in inflamed synovium (B). Bar = 100 μm.

VEGF immunoreactivity.

Cellular VEGF immunoreactivity was detected in all samples of synovium obtained from patients with OA and increased with increasing EC proliferation index, EC fractional area, inflammation grade, and macrophage fractional area (Table 1). In the synovial lining, VEGF immunoreactivity was predominantly localized to CD14-immunoreactive macrophages. Some prolyl 4-hydroxylase–immunoreactive fibroblasts also displayed VEGF immunoreactivity (Figures 1E–H).

Comparison between patient groups.

The distributions of vascular, angiogenic, and inflammation indices and VEGF immunoreactivity were similar in synovial samples obtained from knees and those obtained from hips. Patient age and patient sex showed no relationship to these characteristics. Total osteophyte score (median 8, range 0–12) was positively correlated with JSN score (median 4, range 0–6) in patients with OA (r = 0.54, P < 0.05). No significant associations were found between either radiographic parameter and any of the angiogenic, vascular, or inflammation indices or VEGF immunoreactivity.


Severe synovial inflammation was observed in 31% of patients with OA. Severely inflamed synovial tissue obtained from patients with OA also contained lymphoid aggregates, which often are considered diagnostic of inflammatory joint disease. Synovial inflammation was not confined to patients with extensive radiographic joint damage or end-stage disease. Further studies are required to determine the importance of angiogenesis and inflammation in early OA and their possible contribution to progressive joint damage.

Our data suggest an association between macrophage infiltration and synovial angiogenesis in OA. Macrophages were the predominant VEGF-expressing cells in synovial tissue obtained from patients with OA. The antibody used in our study detects all common VEGF isoforms and provides an indication of the total expression of this angiogenic growth factor within the synovium. In a recent study, it was demonstrated that VEGF expression increased with increasing synovitis in patients with internal derangement of the temporomandibular joint (13). Production of VEGF by synovial macrophages is a possible molecular mechanism exacerbating synovial angiogenesis and inflammation in OA.

In conclusion, we propose that macrophages may mediate synovitis and angiogenesis in OA, and that VEGF may have an important role in both processes. The role of angiogenesis and inflammation in symptoms and disease progression of OA warrants further study.


We are grateful to all patients and the orthopedic surgeons at the King's Mill Centre for Healthcare Services for providing clinical material. We thank the histopathology personnel at the King's Mill Hospital for their help with processing of tissue samples, and Evan Kabir (University of Nottingham) for his contribution to the immunohistochemical analysis.