Crohn’s disease (CD) is a type of inflammatory bowel disease (IBD) characterized by prolonged and chronic ulcerative inflammation that can affect any part of the gastrointestinal (GI) tract. Inflammation in CD is believed to be induced by an excessive and inappropriate immune response against the gut microbiota in genetically susceptible individuals (1). Disease pathogenesis is considered to involve a complex interplay of genetic- and environmental factors, gut microbiota, and host immunity (2, 3). Anti-tumor necrosis factor (TNF) agents such as infliximab (IFX) and adalimumab (ADA) have become the mainstay of treatment for patients with moderate-to-severe CD and have proven effective in induction and maintenance of remission (4, 5). However, up to 30% of patients with CD demonstrate non-response or loss of response after induction therapy, leading to longstanding disease activity followed by complications such as fibrotic strictures and penetrating fistulas (6, 7). Although the aim is always to alleviate these disease complications, approximately 70–90% of patients eventually undergo intestinal resection (8). In clinical practice, it remains challenging to predict which patients will respond to anti-TNF induction therapy. Therefore, biomarkers that may aid in predicting response are urgently needed to improve disease management.

The intestinal tissue is rich in extracellular matrix (ECM) proteins such as collagens, which serve an important structural function (9). The intestinal ECM can be divided into two layers: basement membrane (BM) and interstitial matrix (IM). The BM underlies the epithelium and plays an essential role in epithelial barrier integrity, with type IV collagen being the major collagen component – a network-forming collagen that functions as a barrier between tissue compartments. The IM represents the deeper tissue layers and mainly consists of type I and III collagens, both of which are fibrillar collagens that participate in processes such as wound healing, where they act as a scaffold for fibroblast attachment. Type VI collagen can be found in the interface between the BM and IM, separating the two layers. In CD, the ECM remodeling equilibrium is disturbed due to high protease activity, causing injury to the tissue. This results in an influx of inflammatory cells causing BM degradation, which in turn initiates repair processes with increased ECM formation, feeding into the loop of chronic inflammation (10, 11). Proteases involved in ECM remodeling can be expressed by a variety of tissue-resident cells as well as immune cells. For example, granulocytes and macrophages produce matrix metalloproteinases (MMPs) which are known to degrade ECM components, and T-lymphocytes secrete Granzyme-B (GrzB) that enables them to infiltrate through the BM (12, 13).

Fragments derived from intestinal ECM remodeling due to chronic inflammation are released into the circulation and can serve as biomarkers reflecting the pathological disease process (14). These markers could provide personalized treatment and identify changes in the intestinal tissue on a molecular level, adding on another layer of knowledge. Previous studies reported that specific fragments of type III and IV collagen formation and degradation are associated with CD and hold promise as tools for predicting response to anti-TNF therapy (15–17). Additionally, fragments of type I and VI collagen have shown to be elevated in CD patients compared with healthy controls (15, 18). These biomarkers have been validated by various proteomic techniques, as well as in several autoimmune diseases (19–21).

Here we aimed to provide a proof of concept, showing that serological biomarkers of collagen formation and degradation could monitor or predict response to infliximab induction treatment in patients with CD.

Determining the optimal treatment for CD in a timely and non-invasive manner remains a challenge in clinical practice. In this study, we show that serological biomarkers of collagen formation and degradation have promising potential for predicting and monitoring early response to infliximab treatment in patients with CD, which largely depends on patients’ surgical history. We observed that compared with non-responders, patients with a surgical history that responded to IFX treatment demonstrated elevated baseline levels of type IV (C4M) collagen degradation, as well as elevated levels of type VI (C6Ma3) collagen degradation and formation of type III (PRO-C3), and VI collagen (PRO-C6) at week 14 of treatment. Baseline C4M was associated with response to IFX, while C6Ma3, PRO-C3, and PRO-C6 could distinguish between responders and non-responders at week 14. In the group without surgical history, responders to IFX had lower levels of type IV collagen turnover ratios (C4M/C4G, PRO-C4/C4G). These biomarker ratios could also differentiate between responders and non-responders to IFX at week 14. Moreover, we found that clinical laboratory parameters collected at baseline did not seem to be associated with response to treatment.

When looking at patients with surgical history, we noted that the majority had either stricturing or penetrating disease, or in other words, a more complicated disease phenotype. Stricturing disease is characterized by fibrosis, caused by continuous tissue injury and excess ECM deposition, while penetrating disease is characterized by transmural inflammation. It is important to note that patients with penetrating disease usually also present with fibrosis, meaning that in most cases their disease is not confined to an inflammatory phenotype (22–24). In contrast, patients without surgical history mainly present with a luminal disease phenotype or a less complicated phenotype characterized by superficial inflammation of the GI tract without other complications such as strictures or fistulas (25). We observed significantly elevated levels of type I, III, IV, and VI collagen degradation in patients without surgical history, which could be explained by the phenotypic difference just presented between the two groups. In other words, patients with a less complicated disease phenotype, mainly represented by luminal inflammation, have overall higher levels of ECM degradation compared with patients with a more complicated disease phenotype. This has been observed before in a study conducted by van Haaften et al., where patients with CD were grouped by disease behavior according to the Montreal classification (non-stricturing/non-penetrating, stricturing, or penetrating) and biomarkers of collagen formation and degradation were measured. Results showed that the ratio of type III collagen turnover (PRO-C3/C3M) was higher in non-stricturing/non-penetrating disease vs. stricturing and penetrating disease, implying greater ECM turnover activity within the luminal disease phenotype (15).

Serum levels of type IV collagen degradation (C4M) were elevated at baseline in patients with surgical history who eventually responded to IFX treatment at week 14, compared with those who did not respond. A similar trend was observed for several other collagen degradation markers (i.e., C1M, C3M, and C6Ma3). The ability to separate responders from non-responders at baseline, before treatment initiation, is important in clinical practice as it can aid in patient selection and therapeutic decision-making. Given that this patient group represents a more fibrotic phenotype, these results could point to fibrosis resolution in the BM and IM (elevated collagen degradation) within the patients responding to treatment, as effective fibrosis resolution requires degradation and removal of the fibrotic ECM (26). It might be noted that C4M levels are already elevated at baseline. A possible explanation for this could be that responders have increased fibrolytic activity compared with non-responders. One can also think of the two response groups in terms of “cold” and “hot” fibrosis: “Cold” fibrosis represents fibrotic tissue with little to no ECM remodeling activity, thus no fibrolysis, while “hot” fibrosis corresponds to fibrotic tissue with active fibrolysis capable of carrying out normal wound healing. When comparing serum biomarker levels between patients with surgical history vs. those without, another trend was observed with C4M and a few other collagen degradation markers (C1M, C3M, and C6Ma3). Patients with surgical history generally tended to have lower levels of these markers at baseline and after IFX treatment at week 14. This could be explained by the fact that these patients may have had the most prominent fibrotic part of their intestines removed, while patients without surgical history might have some subclinical fibrosis being resolved, leading to elevated levels of collagen degradation. Moreover, these patients will on average have more viable intestinal (fibrotic) tissue compared with patients with surgical history, likely resulting in higher levels of collagen degradation markers. Using these markers to differentiate responders from non-responders after the initiation of treatment provides important clinical value. The neo-epitopes being detected are directly linked to tissue turnover within the intestinal mucosa. Therefore, the biomarkers could be used to monitor tissue healing or the increase in inflammation and fibrosis, offering potential for, i.e., identifying patients whose fibrosis might not be improving regardless of their other symptoms, guiding treatment escalation or a switch to another biologic. This form of monitoring is also both more objective than clinical disease scores, and less invasive than current endoscopic procedures.

In this context, it is worth mentioning the type III collagen formation marker: PRO-C3. We observed elevated levels of PRO-C3 at week 14 in the responders within the group of patients with surgical history compared with non-responders. The fragment being detected is the neo-epitope of the released N-terminal pro-peptide of type III collagen, meaning that it is a marker of true type III collagen formation. It is widely used within the field of liver fibrosis, as it reflects active liver fibrogenesis (27). A healthy reference range has been established for PRO-C3 with a lower limit of 6.1 ng/ml [6.1–6.1] and an upper limit of 14.7 ng/ml [14.0–15.3] (28). In the study at hand, we observed that both responders and non-responders fall within the healthy range, with responders showing a 31% increase in the biomarker at week 14. This observation suggests an active healing response in patients responding to IFX treatment, and may support the hypothesis of “cold” and “hot” fibrosis. These results are also consistent with the literature, which states that as soon as the cause of tissue injury has been eradicated (i.e., with treatment), ECM remodeling proceeds as part of tissue healing and eventually leads to fibrosis resolution (26). Although PRO-C3 is a known marker for liver fibrosis, increased PRO-C3 levels at week 14 from baseline in responders do not resemble a pathological state. Therefore, in the context of IBD, PRO-C3 could potentially serve as a marker for tissue healing or early fibrogenesis as tissue healing and fibrosis are intertwined. Additionally, it has been shown that PRO-C3 is not able to differentiate CD patients with stricturing disease from other Montreal disease behavior phenotypes, suggesting that it is not a feasible marker for fibrosis in CD (15). A similar pattern was seen with PRO-C6, where within the group of patients with surgical history, responders at week 14 had higher levels than non-responders. PRO-C6 is a C-terminal released C5 domain of type VI collagen α3 chain, also known as endotrophin, typically reflecting pro-fibrotic signaling, and it has been shown to be associated with fibrosis stages in non-alcoholic fatty liver disease (NAFLD) (29). It has been reported that patients with CD in clinical remission (HBI <5) have higher serum PRO-C6 levels than patients with active disease (30). Based on these findings, and the results presented in this paper, it can be hypothesized that PRO-C6 reflects a tissue healing response in patients with surgical history responding to IFX treatment. Furthermore, serum levels of type VI collagen degradation (C6Ma3) were elevated in responders at week 14 compared with non-responders, further strengthening the hypothesis that responders to IFX have active ECM remodeling within their fibrotic tissue as a result of proper wound healing.

Two distinct type IV collagen ratios (C4M/C4G and PRO-C4/C4G) were elevated at week 14 in non-responders within the group of patients without surgical history, compared with responders. C4M/C4G is a ratio between serum concentrations of two type IV collagen fragments: one fragment generated by MMP-2, -9, and -12 (C4M), and the other generated by the serine protease Granzyme B (C4G). MMPs are expressed by many different cell types such as granulocytes, monocytes, fibroblasts, and endothelial cells. For example, MMP-9 is known to be the major secretion product of activated monocytes and a component of the specific granules of human neutrophils, while macrophages are known to be the primary cellular source of MMP-12 (31–33). Granzyme B is believed to be exclusively produced by cytotoxic T-cells and NK-cells. It induces the breakdown of type IV collagen, enabling infiltration through the BM (13). It can be concluded that although these two fragments are derived from the same protein, they represent two different forms of inflammation; C4G can be directly related to T-cell activity and infiltration (lymphoid lineage), while C4M is more representative of granulocyte- and monocyte-mediated inflammation (myeloid lineage). By combining these two biomarkers into a ratio, we aimed to gain insight into differing inflammation profiles in relation to IFX treatment. Our results showed that non-responders had higher C4M/C4G ratios at week 14 compared with responders. Therefore, it can be assumed that non-responders have a higher degree of MMP-mediated type IV collagen degradation (C4M) and lower T-cell activity (C4G). In contrast, responders to IFX have lower levels of C4M compared to C4G, meaning they have less myeloid cell-mediated inflammation and greater lymphoid (e.g., T-cell) activity. Similarly, non-responders showed elevated levels of the ratio PRO-C4/C4G at week 14 of therapy, compared with responders, suggesting that non-responders have relatively higher levels of PRO-C4 compared with C4G. PRO-C4 measures an internal fragment of the 7S domain of type IV collagen and is reflective of BM turnover – which may imply that non-responders have higher BM turnover, but lower T-cell activity. These results partially agree with a previous study conducted on patients with CD, where serum levels of collagen turnover were compared between responders and non-responders to anti-TNF therapy, excluding patients with surgical history. The results showed that PRO-C4 and C4M levels were increased in non-responders at week 8, indicating altered type IV collagen turnover and increased MMP-mediated degradation in patients who eventually experienced non-response to anti-TNF treatment (17). However, a note of caution is due when comparing results. In the current study, 62% of patients were in remission at baseline according to the Harvey-Bradshaw Index (HBI), while in the aforementioned study all patients had clinically active disease at baseline (moderate/severe disease). This disease inactivity might affect the change in biomarker levels over time between responders and non-responders. Patients with mild disease do not significantly differ in marker levels at baseline, yet the non-responders progress in their disease over time. Meanwhile, patients with moderate/severe disease already present with differing biomarker levels at baseline and maintain their response profile over time.

Taken together, responders within the group of patients with surgical history show elevated levels of ECM degradation and formation, indicating fibrosis resolution and tissue healing, which is in harmony with their fibrotic phenotype. These findings are particularly relevant and novel in that patients with CD having prior surgical history are often not investigated independently, i.e., in biomarker discovery studies and clinical trials. Because of the nature of their disease, however, their response profile might look different from patients who have not undergone surgery, and it may therefore be critical to stratify for surgical background. Non-responders within the group of patients without surgical history, many of which were characterized by a luminal disease phenotype, showed elevated levels of biomarker ratios reflecting collagen turnover and degradation of the intestinal mucosa, which may be considered indicative of increased inflammation. This hypothesis is depicted in a flowchart (Figure 5).

Several limitations to this study warrant recognition. First, our sample size was limited, specifically in relation to the group of non-responders, which made it challenging to compare the two response groups. However, this issue was inherent to the study design as the number of responders to IFX generally outweighs that of non-responders with regard to induction therapy. Subsequently, our analyses were stratified by history of prior surgery, which resulted in small subgroups and a decrease in statistical power. A second limitation is the limited availability of disease activity assessment based on the HBI score, both at baseline and week 14. In this study, 30% of HBI scores at baseline were not available, which limited both the assessment of clinical disease activity and the change in activity from baseline to week 14 in relation to the biomarker levels. Furthermore, clinical response to treatment at week 14 was based on a composite of HBI scores, PGA, and steroid-free status, and only PGA and steroid-free status when HBI scores were missing. In this respect, it would have been better to have clinical disease activity defined solely based on standardized HBI scores to enhance comparability to previously performed studies. Another limitation worth noting is missing information on time from surgery to serological measurements. Since ECM remodeling happens over time, differential time duration since surgery may affect these processes. Lastly, for disease activity assessment, we had to rely on clinical and serological assessment of disease activity (i.e., HBI scores, PGA, and CRP values), as data on fecal calprotectin levels or endoscopic disease activity were insufficiently recorded at time of sampling. For future validation studies, these parameters should ideally be readily available to study associations between collagen biomarkers and biochemical and/or endoscopic definitions of treatment response.

In conclusion, this study demonstrates that baseline levels of a serological biomarker for type IV collagen degradation (C4M) is associated with response to infliximab induction therapy in patients with CD and prior surgical history, and that biomarkers of type III and VI collagen formation, and type VI collagen degradation may be used for monitoring of response at the end of induction therapy in patients with surgical history. Furthermore, biomarker ratios of type IV collagen formation/degradation show potential utility for monitoring treatment response in patients without a surgical history. Our results highlight the potential significance and clinical applicability of serological biomarkers of intestinal collagen turnover for prediction and monitoring of response to infliximab therapy in CD, although future studies are warranted to further validate our findings.

The datasets generated for this study are available upon reasonable request to the corresponding author.

The studies involving human participants were reviewed and approved by the Institutional Review Board (IRB) of the UMCG (IRB no. 08/338). The patients/participants provided their written informed consent to participate in this study.

MA, AB, GD, and JM were involved in conceptualization and study design. GD and JM were responsible for funding acquisition and resources. AB, MA, RL, HD, MV, EF, RW, MK, MP, JM, and GD collected the study data and materials. MA performed the data analysis and data visualization. MA and AB wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the final version of the manuscript to be submitted for publication.

MA, MP, MK, and JM are employees of Nordic Bioscience A/S. MK owns stocks in Nordic Bioscience A/S. GD received research grants from Royal DSM and speaker’s fees from Janssen Pharmaceuticals, Takeda, Pfizer, and Abbvie. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.