Antiangiogenic effects of anti–tumor necrosis factor α therapy with infliximab in psoriatic arthritis




Neovascularization, with an increased number of synovial vessels with a characteristic morphology, seems to contribute to the progression of psoriatic arthritis (PsA). Accordingly, angiogenesis may be an important therapeutic target in PsA. The aim of this study was to analyze the effects of infliximab on angiogenesis in the synovial membrane of patients with PsA who responded to this therapy.


The study group comprised 9 patients with PsA who were selected for the presence of active polyarthritis (including knee synovitis) despite methotrexate therapy. Clinical and biologic evaluations were performed at each visit. Arthroscopy and synovial biopsies were performed at week 0, before infliximab therapy was initiated, and at week 8, after administration of 3 intravenous infusions of infliximab (5 mg/kg). We used immunohistochemistry to identify changes in infiltrating cells and in the angiogenesis modulators αvβ3 integrin, vascular endothelial growth factor (VEGF), angiopoietin 2 (Ang-2), flt-1 (VEGF receptor 1 [VEGFR-1]), kinase insert domain receptor [KDR]/flk-1 (VEGFR-2), and stromal cell–derived factor 1 (SDF-1). Neovascularization was assessed by automated histomorphometry of CD31+ vessels and by measuring αvβ3 expression.


Rapid and significant clinical and biological improvement were observed after treatment in all patients. In the synovium, infliximab therapy induced a significant reduction in macrophages, the CD31+ vascular area, αvβ3+ neovessels/Ulex europaeus agglutinin+ vessels, VEGF and its receptor KDR/flk-1 (VEGFR-2), and SDF-1+ vessels. Expression of flt-1 (VEGFR-1), and SDF-1 in lining cells showed a nonsignificant reduction, whereas expression of Ang-2 increased. In 3 patients, reverse transcription–polymerase chain reaction confirmed the changes in some of these markers at the messenger RNA level.


These results show consistent changes in several factors involved in angiogenesis regulation, in parallel with the clinical response to infliximab in patients with PsA. The pattern of reduced VEGF with increased Ang-2 suggests vascular regression as a potential mechanism underlying the antiangiogenic effect of infliximab.

The pathogenesis of psoriasis involving the skin or joints is not completely understood at present. However, there is evidence that tumor necrosis factor α (TNFα) is involved in the pathophysiology of the disease, at both the skin and synovial level. In psoriatic arthritis (PsA), TNFα activates NF-κB, leading to synovial cell proliferation, leukocyte trafficking, further proinflammatory cytokine production, and up-regulation of RANKL-mediated osteoclastogenesis (1, 2).

Angiogenesis is believed to play a pivotal role in the pathophysiology of synovitis in PsA, from the early stages. Therefore, angiogenesis may be a relevant therapeutic target in PsA (3). Angiogenesis is controlled by specific growth factors, such as vascular endothelial growth factor (VEGF) and angiopoietins Ang-1 and Ang-2, which may be produced in response to cytokines including TNFα. TNFα augments angiogenesis through the expression of VEGF and its endothelial receptor (VEGFR-2; kinase insert domain receptor [KDR]/flk1) (4). In PsA, elevated VEGF concentrations have been reported in serum and synovial fluid (5), and Ang-2 has been detected in synovial tissue (3).

Supporting the concept that TNFα plays an important role in the pathogenesis of PsA, several studies have demonstrated that anti-TNFα therapy is an effective treatment for the skin and joint manifestations of PsA (6). In rheumatoid arthritis (RA), the description of synovial changes induced by anti-TNFα therapy has provided insight into its mechanisms, leading to better identification of cell types and molecules involved in the pathogenesis of RA (7). Although some studies have addressed the clinical effects of anti-TNFα therapy in PsA, only one has analyzed its effects on cellular infiltration and vascularity (8); the potential mediators and mechanisms involved have not been investigated.

The objective of the present study was to analyze the effects of anti-TNFα therapy on angiogenesis in the synovial tissue of patients with PsA, in association with the clinical response to therapy.


Patients and synovial biopsies

Nine patients with PsA were included as a subgroup from an open-label study of the anti-TNFα monoclonal antibody infliximab (the MIPRA study). Patients were treated with intravenous infusions of infliximab (5 mg/kg) at baseline, week 2, and week 6. All patients had active polyarthritis and were selected for a lack of response to methotrexate (15 mg/week) over a period of at least 3 months and by active synovitis (pain and inflammatory synovial fluid) of the knee at baseline. Treatment with methotrexate, nonsteroidal antiinflammatory drugs, and steroids (prednisone, <7.5 mg/day) was continued at stable dosages throughout the study. All patients gave informed consent, and the study was approved by the ethics committee of the Hospital Clinic of Barcelona.

The clinical evaluation at baseline and at week 8 included the number of tender joints of 68 joints counted, the number of swollen joints of 66 joints counted, the erythrocyte sedimentation rate (ESR), and the C-reactive protein (CRP) level. Arthroscopy was performed with a 2.7-mm–diameter arthroscope (Storz, Tuttlingen, Germany), without lavage (for technical reasons, only 150–250 ml of saline was injected), and no intraarticular steroids were injected. Eight synovial biopsy specimens per patient were obtained from the suprapatellar pouch and the medial gutter at baseline and week 8. Biopsy specimens were either fixed in 4% formaldehyde and paraffin embedded or were snap-frozen in OCT compound.

Immunohistochemical analysis

Paraffin sections were subjected to antigen retrieval in a pressure cooker when necessary. Slides were immunostained using an automated immunostainer (TechMate 500 Plus; Dako, Carpinteria, CA). Serial sections were incubated with the following antibodies: anti-CD4 (1F6; Novocastra, Newcastle, UK), 1:50 dilution; anti-CD20 (L26; Dako), 1:800 dilution; anti-CD8 (4b11; Novocastra), 1:10 dilution; anti-CD68 (KP-1; Dako), 1:6,000 dilution; anti-CD138 (B-B4; Santa Cruz, San Diego, CA), 1:10 dilution; anti-CD31 (JC70A; Dako), 1:40 dilution; anti-VEGF (polyclonal goat IgG; R&D Systems, Abingdon, UK), 0.33 μg/ml; anti-VEGFR-1 (flt-1) (polyclonal rabbit IgG; NeoMarkers, Fremont, CA), 10 μg/ml; anti-VEGFR-2 (KDR/flk-1) (polyclonal rabbit IgG; NeoMarkers), 10 μg/ml; and Ang-2 (polyclonal goat IgG; R&D Systems), 10 μg/ml. Primary antibodies were detected by an avidin–biotin–peroxidase–based method (EnVision system, Dako). Peroxidase activity was developed with 0.05% hydrogen peroxide and aminoethylcarbazole (Sigma, St. Louis, MO). Slides were counterstained in hematoxylin. Immunodetection of stromal cell–derived factor 1 (SDF-1), αvβ3 integrin, and Ulex europaeus agglutinin type I (UEA-I) was performed as previously described (9). Preliminary studies in human tonsils, placenta, and angiosarcoma tissue were used to optimize antibody concentrations. Negative controls included mouse or rabbit IgG, or normal goat serum, at the highest concentration of primary antibodies.

Microscopic analysis

All sections were randomly evaluated by 2 independent observers who were blinded to the clinical data. The compensated kappa value of interobserver reliability was 0.85. Immunohistochemical results were scored semiquantitatively on a 0–4 scale, as follows: 0= no reactive cells, 1 = 1–10% positive cells, 2 = 10–25% positive cells, 3 = 25–50% positive cells, and 4 = >50% positive cells.

The vascular area was quantified by CD31 immunostaining with ImagePro software version 3.0 (Olympus Microscopy, Melville, NY). Units are given in micrometers (pixels/unit x = 51.600; pixels/unit y = 51.600), and the total area measured was 161.64 μm2 by field (TV monitor area). Four random 40× power fields were counted in each slide by an independent observer blinded to clinical data. We also analyzed the percent of neovessels as identified by double fluorescent labeling of CD51/61 (αvβ3 integrin) and UEA as a universal endothelial marker.

Reverse transcription–polymerase chain reaction (RT-PCR)

In order to confirm the results of immunohistochemical analysis, RNA was extracted and reverse transcribed. All primers spanned at least one intron, in order to discriminate from genomic contamination. In addition, the selected primers allowed for the detection of alternatively spliced isoforms. The upstream and downstream primers were as follows: VEGF, CTCCGAAACCATGAACTTTCTGCTG and CAGTCTTTCCTGGTGAGAGATCTG; VEGFR-2 (KDR/flk-1), TACCGGGAAACCGACTTGGGCT and TCTTTTCTTGGTCTTCCTGTCTTG; Ang-2, GGAAGAGCATGGACAGCATAGG and CTTGGATACTAACACCTGTAGCTG; Ang-1, GGAGTGTGCTGGCAGTACAATG and TTTCCTGCTGTCCCAGTGTGAC; Tie2, GGTGCCATGGACTTGATCTTGATC and GCAGTACAGAGATGGTTGCATTCAG; and β-actin, GTCACCAACTGGGACGACATG and GCTCGGTGAGGATCTTCATGAG.

Statistical analysis

Scores obtained at baseline and week 8 were compared using the paired Wilcoxon's signed rank test or McNemar's test. P values less than or equal to 0.05 were considered significant.


Clinical response

Nine patients (5 men and 4 women) were included in the study and underwent all procedures without experiencing any complications. The mean age of the patients was 44.8 years (range 29–66 years), and the mean disease duration was 10.1 years (range 2–25 years). All patients had polyarthritis without axial involvement and were rheumatoid factor negative and HLA–B27 negative. At week 8, all patients experienced a marked and significant improvement over baseline in clinical parameters (tender joint count 15.1 ± 8.8 versus 6.0 ± 12.2 [P = 0.017], swollen joint count 9.6 ± 5.4 versus 2.1 ± 2.6 [P = 0.001]) as well as in acute-phase reactants (ESR 47.4 ± 22.9 mm/hour versus 18.8 ± 12.5 mm/hour [P = 0.001], CRP level 3.5 ± 1.8 mg/dl versus 0.5 ± 0.7 mg/dl [P = 0.004]). In addition, none of the patients had knee arthritis on clinical examination at week 8.

Immunohistochemical evaluation

Markers of synovial tissue cellular infiltration

The mean score for macrophages (CD68+) was significantly reduced at week 8 compared with baseline, both in the lining layer (2 ± 1.1 versus 0.9 ± 0.3; P = 0.02) and in the sublining area (1.2 ± 0.7 versus 0.3 ± 0.5; P = 0.02) (Figure 1). The mean scores at week 8 and baseline for CD4 lymphocytes (2.2 ± 1.1 versus 2.8 ± 1), CD8 lymphocytes (1.8 ± 0.9 versus 1.8 ± 0.9), plasma cells (CD138+) (1.9 ± 1 versus 2.1 ± 1.4), and B cells (CD20+) (0.8 ± 1 versus 1 ± 1) were not significantly different.

Figure 1.

Immunohistochemical detection of several cell and angiogenesis markers in representative sections obtained from the same patient with psoriatic arthritis before (week 0) and after (week 8) infliximab therapy. CD68 (macrophages), CD31 (vessels), vascular endothelial growth factor (VEGF), angiopoietin 2 (Ang-2), VEGF receptor 1 (VEGFR-1; flt-1), VEGFR-2 (flk-1/kinase insert domain receptor), and stromal cell–derived factor 1 (SDF-1) (immunoperoxidase stained and hematoxylin counterstained); Ulex europaeus agglutinin (UEA; vessels) (rhodamine stained); and anti-αvβ3 integrin (neovessels) (fluorescein stained). Control (CTRL) sections with rabbit (top) and mouse (bottom) IgG at the highest concentration of primary antibodies. For each marker, the top panel corresponds to a basal biopsy specimen, and the bottom panel corresponds to a biopsy specimen obtained from the same patient after infliximab therapy. Sections are representative of 6–9 patients. (Original magnification × 400.)

Changes in neovascularization

The vascular area showed a significant reduction after infliximab therapy (P = 0.04) (Figure 1 and Table 1). In 7 of 9 patients, the vascular area diminished significantly, whereas in 2 patients it was unchanged. To further confirm neovessel reduction after therapy, we analyzed changes in the percentage of neovessels, identified as αvβ3+, among all UAE+ vessels. The percentage of αvβ3+ vessels decreased after therapy in 6 of 7 patients (P = 0.02) (Figure 1 and Table 1).

Table 1. Immunohistochemical scoring for angiogenesis markers in synovial biopsy specimens*
MarkerWeek 0Week 8P
  • *

    Values are the mean ± SD. Vascular endothelial growth factor (VEGF), angiopoietin 2 (Ang-2), VEGF receptor 1 (VEGFR-1), VEGFR-2, stromal cell–derived factor 1 (SDF-1) lining, and SDF-1 vessels were scored on a 0–4 scale. UEA = Ulex europaeus agglutinin.

VEGF2.4 ± 1.21.7 ± 1.1<0.05
Ang-22.7 ± 0.53.4 ± 0.7<0.05
VEGFR-11.9 ± 1.20.9 ± 0.90.08
VEGFR-22.2 ± 1.31.0 ± 1.1<0.05
SDF-1, lining2.5 ± 1.01.7 ± 0.50.06
SDF-1, vessels2.2 ± 1.50.7 ± 0.5<0.05
CD31 vascular area, μm19.9 ± 14.511.5 ± 15.3<0.05
αvβ3 neovessels, %16.4 ± 6.38.0 ± 7.6<0.05
αvβ3 neovessels/UEA, %27.1 ± 17.310.8 ± 10.2<0.05

Markers of angiogenesis

Expression of VEGF was observed in the lining layer and sublining cells around vessels (Figure 1). The mean score for VEGF expression at week 8 was significantly decreased compared with the mean score at baseline (P = 0.04) (Table 1). In 6 patients the mean score was reduced, in 2 patients there was no change, and in 1 patient it was slightly increased. The mean score for Ang-2 expression showed a significant increase after therapy (P = 0.02) (Table 1). In 7 patients it was increased, and in 2 patients it was unchanged. Ang-2 localization was similar to that of VEGF, but it also included mast cells (Figure 1).

VEGFR-1 (flt-1) displayed diffuse expression in the lining and sublining layers (Figure 1). The mean VEGFR-1 score was reduced after therapy in 7 of 9 patients, and the overall trend approached, but did not reach, statistical significance (P = 0.08) (Table 1). VEGFR-2 (KDR/flk-1) was expressed by perivascular cells and displayed diffuse expression in the sublining layer (Figure 1). The mean VEGFR-2 score diminished significantly at week 8 (P = 0.04). It was reduced in 7 patients, unchanged in 1 patient, and slightly increased in another patient (Table 1).

We also studied expression of the chemokine SDF-1, an endothelial chemoattractant involved in RA angiogenesis (9). The observed pattern of SDF-1 expression was similar to that previously observed in RA synovium, with strong expression by lining and sublining synoviocytes and on endothelium (Figure 1). The mean score for SDF-1 expression in lining and vessels decreased after therapy in 4 of 6 patients and in 5 of 6 patients, respectively. SDF-1 expression was significantly reduced in vessels (P = 0.04) and tended to decrease in the lining (P = 0.06) (Figure 1 and Table 1).


Good quality RNA could be obtained from 3 patients before and after infliximab therapy. After therapy (Figure 2), expression of VEGF121, VEGF165, VEGFR-2 (KDR/flk-1), and Tie2 was decreased. In contrast, the expression of Ang-2 increased after treatment. The Ang-2/VEGF ratio increased 3- to 6-fold after infliximab therapy, in accordance with the data obtained by immunohistochemical analysis in these patients. Ang-1 did not change after therapy.

Figure 2.

Reverse transcription–polymerase chain reaction expression of markers of angiogenesis in synovial tissue before (week 0) and after (week 8) infliximab therapy. Ang-1 = angiopoietin 1; VEGF = vascular endothelial growth factor (isoforms 121 and 165); VEGFR-2 = VEGF receptor 2. Results are representative of 3 patients.


We present the results of a pilot study of infliximab therapy for PsA, focusing on angiogenesis changes as a potentially relevant consequence of this intervention. We performed sequential histologic evaluation of synovial tissue with the purpose of identifying cellular and molecular targets that potentially are modified by this therapy. Our patients were selected on the basis of having active knee synovitis at baseline. As expected from previously reported data in PsA (6), anti-TNFα therapy induced a major clinical response in all of the patients, with a significant improvement in peripheral synovitis, the ESR, and the CRP level. Our immunohistochemical data suggest that a reduction in macrophages and several angiogenesis markers could be a relevant mechanism of action of infliximab in psoriatic synovitis. However, T lymphocyte subsets, B lymphocytes, and plasma cells were unchanged after 8 weeks of therapy. This finding is in concordance with those of previous studies showing a good correlation between macrophage infiltration and clinical signs and symptoms of arthritis. In fact, there are no examples of successful therapy in which synovial macrophage infiltration was unchanged (10).

Our immunohistochemical and RT-PCR data suggest that reduction of synovial angiogenesis after anti-TNFα therapy may be related to modulation of several molecular factors involved in this process, such as VEGF and its receptors (VEGFR-1 and VEGFR-2), the angiogenic chemokine SDF-1, and Tie2 receptor. In contrast, Ang-2 expression was increased after therapy. Recently, it was shown that Tie2 signaling is an important angiogenic mediator that links TNFα to pathologic angiogenesis (11). Ang-2, a natural inhibitor of Tie2, may down-regulate this signaling pathway, inducing vascular regression. Although Ang-2 is an initiator of angiogenesis in the presence of VEGF, this angiopoietin becomes essential for vascular regression when VEGF is down-regulated (12). An increase in the Ang-2/VEGF ratio after infliximab treatment, as well as Tie2 reduction, as confirmed by RT-PCR in 3 of our patients, could explain the consistent reduction of synovial neovascularization observed in our study.

The observed decrease in VEGF expression as a consequence of anti-TNFα treatment can be explained because VEGF expression is partially regulated by NF-κB, a key transcription factor in the signaling pathway of TNFα. Because VEGF is a potent inductor of Ang-2 expression, it is less clear why in this study the expression of Ang-2 increased after anti-TNFα treatment. In addition, it has been reported that TNFα induces the expression of Ang-2 in cultured endothelial cells, and this induction is NF-κB dependent (13). However, this effect has not been observed in other cells types, such as RA synovial fibroblasts (14). Taken together, these findings suggest that TNFα regulation of Ang-2 expression may be more complex than expected.

There is only one previous study on immunohistologic changes in patients with PsA (n = 4) who were treated with infliximab (8). The authors of that study reported that at 12 weeks, infliximab reduced macrophages, VCAM-1, and vascularity, whereas changes in CD4 and CD8 lymphocytes were not observed, and the number of B cells and plasma cells increased. Our study confirms these data, with the exception of B cell and plasma cell changes. Because the methodology is similar in both studies, the differences could relate to different clinical patterns or disease duration of PsA.

Our results are consistent with data obtained from patients with RA. Previous studies of the effects of anti-TNFα therapy on expression of Ang-2, VEGF receptors, or SDF-1 in RA have not been described. Recently, it was shown that in murine arthritis, the inhibition of flt-1 (VEGFR-1) protects against joint destruction by suppressing synovial inflammation and angiogenesis (15). Thus, flt-1 constitutes a potential candidate for therapeutic modulation of angiogenesis in arthritis.

Another angiogenic factor modified by anti-TNFα therapy in our patients was the chemokine SDF-1. Because SDF-1 is not directly induced by TNFα in synoviocytes (9), it represents a potentially independent target for antiangiogenic therapy. Although its up-regulation in inflammatory diseases is not well understood, it is possible that a reduction of synoviocyte hyperplasia may explain the reduced expression of SDF-1 observed after infliximab therapy.

This study is the first to show consistent changes in several factors involved in the regulation of angiogenesis in parallel with clinical and biologic improvement of PsA with infliximab therapy. These findings underline the relevance of angiogenesis in the pathogenesis of chronic PsA synovitis and point toward angiogenesis as a potential therapeutic target.


We thank Mireia Castillo and Vanesa Morente for help in immunohistochemical techniques, Fernando Arenzana-Seisdedos for providing anti–SDF-1 K15C monoclonal antibody, and Montse Pau for technical support on RT-PCR.