Is “CTGF” Still a Viable Anti-Fibrotic Target?

Connective tissue growth factor, or ‘CTGF’, was identified in the early 1990s based on its purported growth factor-like properties [1]. Not long after, ‘CTGF’ was shown to be induced by the potent pro-fibrotic cytokine transforming growth factor-beta (TGFbeta) [2]. Based on these observations, Gary Grotendorst and FibroGen proposed that ‘CTGF’ was a downstream mediator of TGFbeta’s pro-fibrotic effects [3]. One of the target indications was scleroderma [4,5,6]. At FibroGen, two drug development programs were formed: one aimed at blocking either the action of CTGF through an anti-CTGF antibody, the other aimed at blocking the ability of TGFbeta to induce the CTGF promoter. This led to partnerships with FibroGen, with Taisho and Sankyo, respectively. On one hand, this led to the identification of CTGF as a target gene of the mechanosensitive transcription factor YAP1 and contributed to the overall consensus that an autocrine pro-adhesive signaling loop centered on the YAP/hippo pathway was responsible for fibrosis [7,8,9]. On the other hand, this led to the development of FG-3019 (pamrevlumab), an anti-CTGF antibody, and multiple failed clinical trails centered on FG-3019 [10,11]. I have reviewed the first topic, pertaining to scleroderma, elsewhere [12]. This present commentary focuses on the second topic, namely the failure of anti-CTGF strategies to work clinically. The story starts in the late 1990s where it became clear that ‘CTGF’ was not a stand-alone molecule. Instead, based on structural similarity, ‘CTGF’ was a member of what came to be known as the CCN family of matricellular proteins consisting of: CCN1 (cyr61), CCN2 (ctgf), CCN3 (nov) and CCN4-6 (wisp 1-3), which act coordinately [13,14]. These proteins were shown not to have growth factor properties but to be pro-adhesive, based on their abilities to bind a variety of integrins and heparan sulfate proteoglycans. For a summary of these studies, the reader is referred elsewhere [15]. The initial notion that CCN2 was a downstream mediator of the actions of TGFbeta did not have direct evidence when it was initially proposed; rather, CCN2 was shown subsequently by several groups to be a cofactor that augmented the action of TGFbeta in vitro and in vivo [16,17]. In vitro, this activity required the presence of CCN2 in the microenvironment and did not occur in naïve cells [18,19]. Despite these concerns, in the early 2000s, FG-3019 was developed. FG-3019 was selected based on its anti-fibrotic activity in animal models, initially in those relying on the effects of continued applications of exogenously added TGFbeta [20] and subsequently in other models, including those of scleroderma fibrosis [21]. Despite its use in the research community, it is noteworthy that FG-3019 was never been demonstrated to have efficacy in any established in vitro assay, including adhesion or TGFbeta synergy assays. Significantly, its anti-fibrotic effects have not been attributed to its ability to bind to and antagonize CCN2. Conversely, gene knockout approaches have been used to show that CCN2 is essential for fibrosis in a variety of animal model systems; these studies have been used as a basis to demonstrate that CCN2 is a bona fide anti-fibrotic target [22,23,24]. Thus, the failure of an anti-CCN2 drug in the clinic may not relate to the absence of legitimacy of CCN2 as a target (although, please see below), but rather due to issues relating to the drug itself. Crucially, the general absence of published mechanistic studies on FG-3019 makes it difficult to evaluate the drug properly. Notably, sufficient FG-3019 was made available to selected researchers to conduct initial efficacy studies only. However, a key paper published by FibroGen scientists may be insightful [25]. This study showed that FG-3019 is eliminated in the plasma proportionate to the levels of circulating CCN2. The authors state: “In a subject with disease that may be producing and shedding more CTGF, target-mediated elimination would be expected to play a larger role in the clearance of FG-3019….(n)o staining for FG-3019 was seen in the heart and lung of animals treated with FG-3019 only”. These results suggest that FG-3019 poorly penetrates tissue, with circulating CCN2 levels correlating with the degree of fibrosis [23,24]. These data collectively suggest that FG-3019 may inherently be a poor drug. A recent study further emphasized these concerns: FG-3019 showed poor target tissue penetration [26]. Collectively, in addition to speculation that pamrevlumab may have failed due to differences in patient inclusion/exclusion criteria between the Phase II and III idiopathic pulmonary fibrosis (IPF) clinical trials and concerns applicable to IPF trials in general [10,27,28], these results suggest that FG-3019 was simply an suboptimal drug candidate. Having established that CCN2 is a bona fide anti-fibrotic target and FG-3019 may not have been an ideal drug candidate, where does this leave us? One option would be to use a different anti-CCN2 strategy that better penetrates target tissue [25,26]. Another option is to recall what turned out to be a major mistake: the discoverer and marketer of ‘CTGF’ as an anti-fibrotic target either ignored, or refused to consider, that ‘CTGF’ is a member of a multiprotein family, that have redundant functions (for detailed discussions, including illustrations, of these concepts, please see refs [14,23,24]). Indeed, CCN1 and CCN4 both have, depending on the context, pro-fibrotic roles, including in the lung and in a scleroderma skin model [29,30,31,32]. Due to this functional redundancy, drugs targeting CCN2 only may not be effective clinically. FG-3019 was designed based on its ability to target CCN2 alone. Perhaps a superior strategy might be to develop approaches based on other members of the CCN family, such as CCN3, that are anti-fibrotic and would be expected to antagonize all pro-fibrotic CCNs [33,34]. Alternatively, using a combinatorial approach using drugs that target individual CCN proteins may be useful. In any case, mechanistic insights underlying how any future anti-CCN drug works are required. These insights could be gleaned using robust in vitro and in vivo assays, including modern techniques such as spatial and scRNAseq transcriptomic analyses, and validation with clinical samples are required. In summary, the failures of FG-3019 in the clinic are not likely to reflect on the suitability of the CCN family—or even CCN2—as an anti-fibrotic target, but rather in the quality of the drug itself, and the lack of insight into how FG-3019 was selected or functions. Thus, future science-based approaches are required to investigate whether anti-CCN drugs can be used clinically to treat fibrosis.