Serum connective tissue growth factor is a highly discriminatory biomarker for the diagnosis of rheumatoid arthritis

Rheumatoid arthritis (RA), which is the most common chronic inflammatory joint disease that affects approximately 1% of the world’s population [1], is characterized by synovial joint inflammation, progressive joint destruction, and disability [2]. Currently, the clinical diagnosis of RA mainly relies on joint involvement, acute-phase reactants, duration of symptoms, and serological indices, including traditional rheumatoid factor (RF) and the presence of anti-citrullinated protein antibody (ACPA) [3].

However, in clinical use the 2010 criteria remain deficient, especially the serological indices. First, serological indices lack sensitivity or specificity. According to the meta-analysis of Nishimura et al., the sensitivity and specificity of a new serum index ACPA for the diagnosis of RA were 67% and 95%, respectively, and the sensitivity and specificity of the traditional index RF were even lower at 69% and 85% [4], respectively. As ACPA has high specificity, and RF has relatively higher sensitivity, the current recommendation is to combine RF and ACPA to detect RA. This combination truly improves the diagnostic value of these tests. However, studies have found that even the combination of these two markers is not perfect, with sensitivity of 78% and specificity of 82% [5] to detect RF-positive or ACPA-positive patients.

The second problem is the poor detection of early RA (ERA) by ACPA, with sensitivity as low as 57%. Similarly, the combination of ACPA and RF has limited performance benefits over either individual index [6, 7]. Therefore, the search for new serum biomarkers, especially those with high specificity and sensitivity to improve the current diagnostic tests for RA, retains great significance.

Connective tissue growth factor (CTGF) was first discovered by Bradham in 1991 [8] and belongs to the CCN family of growth factors, named CCN2. It is a 38 kD cysteine-rich protein made up of four domains, including insulin-like growth factor binding protein (IGFBP), von Willebrand factor type C repeat (VWC), thrombospondin type 1 repeat (TSP1), and C-terminal cystine-knot (CT) modules [9]. CTGF plays an important role in many physiological and pathological activities [10], such as inflammation, angiogenesis, wound healing, fibrosis, carcinogenesis, and tumor development [11]. There is some evidence that CTGF could also be involved in the onset of RA. Nozawa identified increased expression of CTGF in serum from 39 patients with RA when compared to patients with osteoarthritis (OA) and further confirmed that CTGF could enhance the activity of osteoclasts by stimulating integrin protein α_vβ₃ to aggravate bone destruction [12]. Moreover, Fujishiro et al. found that inhibiting CTGF by neutralizing the anti-CTGF monoclonal antibody (mAb) significantly ameliorated arthritis in a murine model of RA [13]. In addition, our previous proteomic study [14] and the subsequent validation tests using PCR and western blot found that CTGF was significantly elevated in the synovial fibroblasts of 50 patients with RA in comparison with 50 healthy controls, raising the possibility that it could be a potential diagnostic biomarker for RA. To evaluate the diagnostic value of CTGF, we performed a multicenter validation cohort study to determine the discriminatory value of CTGF in RA diagnosis.

RA is a chronic autoimmune disease with symmetric polyarthritis as one of its main manifestations. Diagnosis can be challenging, particularly in early disease and in patients with atypical presentation [15]. Thus, much research and effort have gone into the development of RA diagnosis criteria in order to improve the diagnosis and classification of RA. The 1987 ACR criteria for RA, comprising morning stiffness, arthritis, rheumatoid nodules, serum RF, and radiographic changes, were in use for many years [16]. However, these criteria are not very efficient, in that the criteria only help to diagnose those patients who already have serious structural damage. In addition, some of the criteria, like radiographic changes, are not accurate quantitative indicators and mainly depend on the subjective judgment of rheumatologists [17]. These weaknesses stimulated the development of the current gold standard, the 2010 ACR criteria for RA. Compared to the previous version, the new criteria did not include the subjective and inaccurate criterion of radiographic changes, but included a more specific and objective biomarker, ACPA. A systematic literature review shows that the 2010 criteria help to diagnose more patients in the earlier stages of RA. The sensitivity of these criteria has risen by 11%, at the cost of a decline in specificity of only 4% [18].

There remains a need for improved diagnostic tests and criteria in RA, in particular for identifying the disease and to assist in distinguishing RA from other rheumatic diseases. In the current study, we demonstrated that CTGF could partially solve this problem.

One of the most important problems in RA diagnosis is a lack of sensitivity or specificity of the current indicators, ACPA and RF. ACPA is a good clinical indicator of RA (high predictability values and high diagnostic accuracy) and has moderate sensitivity of approximately 67% [4, 19]. However, the sensitivity of ACPA is lower in early RA. RF fares even worse, with a range of specificity from 38% to 85%, indicating that positive IgM-RF has modest diagnostic value [4, 20]. In the current study, we demonstrated that serum CTGF was an excellent serum indicator, with better performance than either of these widely used assays. As an independent biomarker, the sensitivity, specificity, and AUC of CTGF appeared high at 0.86, 0.92, and 0.92, which was much better than for RF, and CTGF had similar specificity but better sensitivity when compared with ACPA. The fact that ACPA is part of the ACR 2010 classification criteria would be expected to inflate its performance to diagnose RA using these criteria, and the relative superiority of serum CTGF in the current study is notable. Furthermore, approximately 30% of patients with RA who have clinical manifestations of the disease may have negative ACPA results [21]. Therefore, we tested the diagnostic ability of CTGF in the ACPA-negative population, and found that serum CTGF was able to identify more patients with RA, with sensitivity and specificity of 76% and 91%, respectively.

The combination of RF and ACPA performs significantly better than either marker alone in the diagnosis of RA [5, 22]. However, researchers also reported that a model including ACPA and RF can correctly identify only 54–57% of patients with RA [22]. In our study, a combination of CTGF, ACPA, and RF had the best diagnostic efficiency, better than single indicators, the combination of RF and CTGF, or even the recommended assay of combined ACPA and RF, with an AUC of 0.97. However, as the AUC of the combination of CTGF and ACPA was not significantly weaker than the combination of the three, we recommend the use of the combination of CTGF and ACPA.

Biomarkers are also of clinical utility in distinguishing RA from other rheumatic diseases. Matsui et al. [23] reported a relatively high frequency of ACPA in patients with other rheumatic diseases, including SLE (15%), pSS (14%), polymyositis/dermatomyositis (23%), and scleroderma (16%). In the present study, we found serum CTGF has good discriminatory capacity in distinguishing RA from other rheumatic diseases, with the AUC for serum CTGF detection greater than 0.92 for all diseases tested except for pSS. The discriminatory capacity of serum CTGF is extremely high in distinguishing SLE from RA and OA from RA.

Synovial proliferation and joint erosion are characteristic features of RA [24, 25]. CTGF has been shown to stimulate hyperproliferation of fibroblast-like synoviocytes in RA [26], and to act as an angiogenesis factor in the formation of pannus. These roles may explain why CTGF discriminates well between RA and other inflammatory arthropathies, and demonstrates face validity for the biomarker.

There were some negative results in our study that might contribute to the application of CTGF in diagnosing RA. No correlation was observed between serum concentrations of CTGF and the duration of RA symptoms. This indicated that the diagnostic efficiency of CTGF may not decrease in early RA, whereas there could be a significant decline in the sensitivity of ACPA when diagnosing patients with ERA rather than patients with established RA [7]. Interestingly, almost no association of CTGF with DAS28 was found in our study, indicating that CTGF might be used without too much consideration of different disease activities.

Synovial fluid detection could also give some clues in distinguishing RA from diseases that are difficult to identify. We quantified CTGF in synovial fluid from patients with RA and the control subjects, and we found that this was a better biomarker than CTGF in serum, with a higher AUC for diagnosing RA. Additionally, the RA-to-control ratio of the concentration of CTGF in synovial fluid and serum are 60 and 10, respectively, and in synovial tissue, according to our previous proteomic study the ratio was 2.54 [14]. Furthermore, there is strong correlation between serum CTGF and synovial fluid CTGF, indicating the same sources of CTGF production. Thus, we surmised that rather than the chondrocytes [27], the synovial tissue or the synovial fibroblasts [12] in the tissue was the initial source of CTGF, and CTGF was then secreted into the synovial fluid, participating in pannus formation as an angiogenesis factor [28], and later diffused into the blood. This hypothesis needs further validation.

Our study has several limitations. Although we have carried out a multicenter study, our participating centers were all from eastern or northern areas of China, and therefore further studies in different ethnic groups are warranted. We also have not tested the performance of CTGF in a specific cohort of patients with early arthritis, nor in preclinical disease, which are situations where ACPA has demonstrated diagnostic or predictive utility. Further testing of the impact of treatment on CTGF levels is also indicated.

In summary, in a multicenter study we identified an excellent and stable biomarker for RA. Serum CTGF concentration has much better diagnostic ability than RF, and is not weaker than ACPA. The combination of CTGF and ACPA could further improve diagnostic efficiency. In addition, it is a better biomarker to distinguish RA from other rheumatic diseases. Thus, we recommend CTGF for clinical use in diagnosing RA and distinguishing RA from other rheumatic diseases.