The present study showed the next step towards intra-operative fluorescence-guided surgery for endometriosis. Potential biomarkers to use as a target for fluorescent imaging, previously selected using transcriptomic analysis, were validated to assess protein expression in endometriosis and relevant adjacent tissue using immunohistochemistry. MMP11 showed the largest overexpression in glands compared to adjacent tissue, and additionally showed stromal expression. VCAN showed the largest overexpression in stroma. Both biomarkers showed a beneficial staining pattern, meaning membranous or extracellular.

In our analysis, MMP11 showed beneficial characteristics for intra-operative visualization. It showed a large upregulation in endometrial glands compared to adjacent tissues, combined with a membranous staining pattern. Stroma showed large upregulation with an extracellular staining pattern.

MMP11 encodes for an enzyme, that cleaves alpha 1-proteinase inhibitor but weakly degrades structural proteins of the extracellular matrix [15]. MMPs are involved in the breakdown of extracellular matrix in normal physiological processes, such as embryonic development, reproduction, and tissue remodeling [16]. MMPs, and MMP11 specifically, were expressed in several types of endometriosis and showed a higher expression level in ectopic endometrium compared to eutopic endometrium. [17]. Next to that, MMPs are also found to be involved in the decidualization of endometrial stroma [18,19,20]. MMP11 scored on the TASC score in our previous study due to literature showing diffuse upregulation of MMP11 in patients and a beneficial subcellular location [11].

Interestingly, MMP11 was anticipated to be extracellularly but exhibited distinct membranous staining in glands, additional to an extracellular pattern in stroma. The membranous location was supported by information from datasheets from other MMP11 antibodies and IHC analyses [21, 22]. Despite MMP11 not being classified as a membrane-type MMP, staining might suggest that post-translational modifications, such as glycosylation or phosphorylation, have influenced protein properties [23]. (Pre-)clinical studies are needed to show were MMP11 is visualized using a fluorescent tracer.

Focusing on medication use, upregulation was slightly stronger for patients with hormonal medication. The influence of hormones on MMPs is supported by literature showing that progesterone inhibition of endometrial MMPs during the secretory phase might be deficient in women with endometriosis [20] and that MMPs are influenced by changes in steroid hormone concentration levels [24].

Concentrating on endometriosis type, upregulation was slightly larger for DE. The larger upregulation of MMP11 in DE is supported by MMPs important role in fibrosis [25]. Fibrosis is thought to play an important role in PE, as well as an even more distinct role in DE [3], being the most fibrotic and infiltrated type of endometriosis [26]. This supports the presence of MMP11 in both subtypes of endometriosis, with a slightly higher upregulation of MMP11 in DE. Although small differences were observed in these sub analyses, upregulation of endometriosis was large in all subgroups. This indicates strong potential for MMP11 to use as a universal target for visualization of endometrial glands and stroma for endometriosis surgery.

In addition to MMP11, VCAN showed promising characteristics. It showed large upregulation in stroma compared to adjacent tissue, with an extracellular staining pattern. No relevant difference was seen within hormonal medication use or types of endometriosis, making VCAN a universal biomarker to use as a target for visualization of endometrial stroma.

VCAN might play a role in intercellular signaling and connecting cells with extracellular matrix [27]. It participates in cell adhesion and angiogenesis [28], it induces inflammation and is found to be upregulated in peritoneum of women with endometriosis compared to women without endometriosis [29]. Further, little is known regarding the correlation between VCAN and endometriosis. VCAN resulted in a high TASC score due to gene upregulation in both peritoneal and deep endometriosis, its association with fibrosis and low RNA expression in surrounding tissue [11].

For both MMP11 and VCAN, staining was observed in adjacent tissue (MMP11: vessel walls, VCAN: smooth muscle cells, loose connective tissue, vessel walls), however significant less than in endometriosis, resulting in high TIS differences. Human Protein Atlas shows on single cell level that vascular smooth muscle cells and endothelial cells show RNA expression for VCAN [30], and could therefore indicate specific staining. For MMP11 no RNA expression was observed [31] and no information was available for non-vascular smooth muscle cells and loose connective tissue. Protein expression is observed for VCAN in smooth muscle cells, connective tissue of bowel, bladder and vascular wall [32]. MMPs have shown a role in vascular remodeling [33], are secreted by smooth muscle cells [34], and might also be found in connective tissue [35]. This might suggest that staining in adjacent tissue could be specific. Further research with fluorescent tracers should focus on specificity of staining in adjacent tissue.

Although our previous analysis only selected biomarkers with a low RNA expression level (expressed in Transcripts per Million (TPM)) for fallopian tubes and bowel [11], both tubal and bowel mucosal epithelium stained positive for VCAN and MMP11. The used TPM values were based on tissue as a whole, being an average of all single cell types. For both MMP11 and VCAN, RNA expression on single cell level showed low TPM values [30, 31]. Therefore, staining might be non-specific. However, with clinical fluorescence use, mucosal staining is anticipated not to pose a significant issue since the bowel and tube will intra-operatively be observed from a serosal side of the organs.

All other analyzed biomarkers showed less favorable combinations of characteristics. Although IGFBP1, CHD2, MMP3 and MMP7 exhibit a high expression compared to adjacent tissue, staining showed a cytoplasmic staining pattern. This suggests less favorable characteristics for the use as a target, as this would result in the need for internalization in the cell. Additionally, cytoplasmic location for expected extracellular located biomarkers could be due to cellular processing and secretion as it may be synthesized in the cytoplasm and then transported to the extracellular space. When the protein is actively secreted or undergoes cellular processing before being released, the cytoplasmic form might be more abundant and detectable by the antibody. Therefore some biomarkers might still have an extracellular location which was not observed in this analysis. Considering this, these biomarkers are not considered the most potential, but should not be completely excluded from future research.

MMP10, PAEP and IL1B revealed low expression in endometriosis, while showing a positive staining in positive control tissue, and are therefore not considered potential for further research. Although CXCL8 showed high expression in endometriosis, all adjacent tissue showed intense staining as well. This may arise from non-specific staining, however given its interleukin nature, this biomarker may manifest ubiquitous high expression. Future research could also focus on the potential of CXCL8.

Previous analysis identified 29 potential targets [11], of which 10 were analyzed in this study. In this previous analysis, a new promising method was used to broaden the identification of candidate targets, combining transcriptomic analysis of publicly available data, with target selection criteria. The current study showed that this method, indeed, resulted in potential targets for FGS. However, multiple selected potential targets, including highly ranked targets, showed less favorable characteristics after protein expression validation. This underscores that RNA expression does not invariably translate directly into protein expression [36]. Still, the used method is considered valuable, as it could result in new potential targets, which would not have been identified with conventional methods [11]. However, validation of potential targets for protein expression is essential.

FGS is thought to be a promising surgical modality for endometriosis surgery. Most new laparoscopy and robotic devices have a fluorescence function. Targeted FGS in endometriosis could be used for several purposes. Mainly, it could be used to enhance identification of endometriosis spots. Additionally, it could be used to determine resection planes by improved identification, to obtain complete excision with enhanced precision and safety. In this study, we focused on visualization of endometriosis itself, however also taking association of the studied biomarker with fibrosis into account [11]. Fibrosis is present in both peritoneal and to more extent in deep endometriosis and develops in reaction to the presence of endometriosis due to recurrent injury and repair [37]. Especially in deep endometriosis, this fibrosis can also be a cause of pain symptoms. However, in contrast to endometriosis, the visualization of fibrosis is considered to be a less significant problem during surgery, as conventional white-light visualization and especially haptic feedback provides valuable information regarding the location. Therefore, by using FGS in combination with white-light visualization and haptic feedback, a more complete and precise removal of endometriosis and associated fibrosis could potentially be achieved.

Studying the tissues showed insightful characteristics of endometriosis, with a ‘patchy’ pattern throughout the tissue. Therefore it could be challenging to visualize endometriosis intra-operatively, as this may result in scattering of the light. Future studies need to focus on clinical behavior of endometriosis during FGS. Additionally, in some nodules, only glands or stromal tissue might be present, which makes concurrent targeting of these structures valuable. Furthermore, glands constitute a small component of the nodule, which points out the importance of stromal visualization. A fluorescent tracer targeting a biomarker staining both glands and stroma, or a dual tracer is thought to be the optimal way for intra-operative visualization of endometriosis.

The potential targets identified in this study, together with previously identified FOLR1 [11, 12], should be analyzed further preclinically by using fluorescent tracers. Pre-clinical models of endometriosis are scarce and challenging. Increasing the predictive value of preclinical models is a hurdle for endometriosis scientists [38]. Research is ongoing on new options to reconstruct human endometriosis to improve translational research [38]. Post-operative ex-vivo endometriosis tissue could be used to validate fluorescent tracers.

Unfortunately, no clinical tracers for both MMP11 and VCAN are available yet. However, numerous fluorescent-labelled or radio-labelled MMP inhibitors for other MMPs have been developed as imaging agents for clinical trials, but additional research is required [39]. However, this knowledge might accelerate the development of a clinical tracer for MMP11. For VCAN, research in this field is ongoing as targeting VCAN is considered a therapeutic potential for multiple diseases [40].

In this study, a large and various cohort of endometriosis samples and healthy specimens was collected. Additionally, the healthy specimens were collected from the same patients. This makes it optimally translatable to clinical use, as the intra-operative differentiation between healthy and endometriosis tissue needs to be made within the same patient. Secondly, multiple biomarkers were tested based on previous selection [11], which resulted in potential targets but concurrently validates the previous new method, using publicly available data, which offers future research potential for other diseases.

As mentioned previously, our analysis revealed occasionally unexpected observed staining patterns and some non-specific staining. As a result, some biomarkers exhibited non-beneficial characteristics based on these observations. However, these biomarkers might still have a potential to use as a target, and could be analyzed further if MMP11 and VCAN prove unsuitable in further research steps towards a fluorescent tracer.