Systematic study of tissue factor expression in solid tumors

Abstract Background Elevated tissue factor (TF) expression, although restricted in normal tissue, has been reported in multiple solid cancers, and expression has been associated with poor prognosis. This manuscript compares TF expression across various solid tumor types via immunohistochemistry in a single study, which has not been performed previously. Aims To increase insight in the prevalence and cellular localization of TF expression across solid cancer types, we performed a detailed and systematic analysis of TF expression in tumor tissue obtained from patients with ovarian, esophageal, bladder, cervical, endometrial, pancreatic, prostate, colon, breast, non‐small cell lung cancer (NSCLC), head and neck squamous cell carcinoma (HNSCC), and glioblastoma. The spatial and temporal variation of TF expression was analyzed over time and upon disease progression in patient‐matched biopsies taken at different timepoints. In addition, TF expression in patient‐matched primary tumor and metastatic lesions was also analyzed. Methods and Results TF expression was detected via immunohistochemistry (IHC) using a validated TF‐specific antibody. TF was expressed in all cancer types tested, with highest prevalence in pancreatic cancer, cervical cancer, colon cancer, glioblastoma, HNSCC, and NSCLC, and lowest in breast cancer. Staining was predominantly membranous in pancreatic, cervical, and HNSCC, and cytoplasmic in glioblastoma and bladder cancer. In general, expression was consistent between biopsies obtained from the same patient over time, although variability was observed for individual patients. NSCLC biopsies of primary tumor and matched lymph node metastases showed no clear difference in TF expression overall, although individual patient changes were observed. Conclusion This study shows that TF is expressed across a broad range of solid cancer types, and expression is present upon tumor dissemination and over the course of treatment.

Conclusion: This study shows that TF is expressed across a broad range of solid cancer types, and expression is present upon tumor dissemination and over the course of treatment. is a transmembrane glycoprotein that is the main physiologic initiator of coagulation when exposed to blood after injury. 1,2 TF also aids in wound healing by stimulating angiogenesis via the protease-activated receptor-2 (PAR2, F2RL1)-mediated intracellular signaling pathway and hampers apoptosis via the Janus kinase (JAK)/signal transducer and activator of transcription (STAT) pathway. 3 TF plays a critical role in embryonic development. 4 Expression of TF under physiologic conditions is limited to some tissues, 5 is predominantly found perivascular, and has been shown to be anatomically sequestered from blood. 5,6 Aberrant expression of TF in cancer was reported over four decades ago, 7 and has been described in a wide range of solid tumors, including breast, ovarian, prostate, pancreatic, bladder, cervical, esophageal, and colon cancer, HNSCC, NSCLC, and glioblastoma compared to normal tissue 8 (Table 1). TF expression in cancer is hypothesized to be induced by several mechanisms including loss of tumor suppressor genes (phosphatase and tensin homolog [PTEN] or p53), activation of oncogenes (e.g., K-RAS and epidermal growth factor receptor variant III [EGFRvIII]), hypoxic tumor microenvironment, or transforming growth factor β [TGFβ] signaling. 9 Mutations in, or amplification of the TF gene itself have not been described. In cancer, TF is thought to facilitate primary tumor growth, neo-angiogenesis, tumor invasion, and metastasis, 10 and TF expression consequently has been associated with poor prognosis in multiple solid cancers. 9,[11][12][13] Expression across solid cancers makes TF an interesting therapeutic cancer target, and preclinical proof-of-concept studies have been described for TF-targeting antibodies, antibody-drug conjugates, immune-conjugates, micro-RNAs, and TF pathway inhibitors. 9,14,15 Comparison of reported TF expression levels between malignancies is difficult, because methodologies used to assess TF expression across individual studies are highly variable, with most studies focused on a single cancer type. In addition, little is known about the cellular localization of TF within tumor cells, the expression pattern in matched primary tumors versus metastatic lesions, or the dynamics of TF expression during disease progression.
In the present study, we set out to determine the prevalence, cellular localization, and the spatial and temporal expression patterns of TF in a broad panel of solid tumor biopsies, taken at various time points and stages of disease, and from primary and metastatic lesions, using validated reagents and IHC methods.

| Kidney tissue sample
Freshly excised normal kidney tissue (VUmc) was directly subjected to standard formalin fixation and paraffin embedding. TF expression was assessed at sequential time intervals ranging from 0-10 months on freshly cut slides, using the validated TF IHC protocol (Method 1, with a primary antibody concentration of 2.5 μg/ml HTF-1).

| Correlation between IHC method 1 and 2
The methodological differences in the two IHC methods are summarized in Table S4. Reproducibility and assay dynamic range of both methods were verified using formalin-fixed paraffin-embedded (FFPE) tumor cell lines with known TF expression. The IHC methods showed high reproducibility and a comparable assay range ( Figure S2).

| Immunohistochemical staining of TMAs and normal kidney biopsies
All tumor TMAs and normal kidney biopsies were stained according to Ventana's TF IHC protocol, which is similar to IHC method 1, with minor adjustments: incubation with CC1 buffer was performed for 64 minutes instead of 36 minutes, incubation with mouse anti-human TF antibody was performed with 2.5 μg/ml instead of 3 μg/ml and staining was performed on a Ventana BenchMark slide stainer.

| Immunohistochemical staining of tumor biopsies
Freshly cut FFPE tumor tissue sections were stained according to IHC staining method 1 or 2, as indicated above, which were verified for reproducibility and assay range ( Figure S2). To exclude potential impact of storage of FFPE tissue blocks on TF protein/epitope stability, it was confirmed via IHC that there was no change in TF staining in an archival FFPE tissue block of healthy kidney tissue for up to at least 10 months ( Figure S3).

| TF mRNA expression in solid tumors
The  Table 1).
In the majority of cancers, TF expression was observed on the plasma membrane, in addition to cytoplasmic staining in some cancer types ( Figure 1). TF staining was predominantly cytoplasmic in bladder cancer and glioblastoma, with no expression in the nuclei. Although only membrane-expressed TF can form TF:FVIIa complexes, it has been shown that TF recycles quickly between the membrane and the cytoplasm. 17 Analysis of TF mRNA expression using The Cancer Genome Atlas (TCGA) database confirmed TF expression across solid cancer types, with heterogeneity between and within cancer types ( Figure 2).
Together, these results show that TF is abundantly expressed across solid tumors, and expression is observed both on the cell membrane and in the cytoplasm of tumor cells.

| Temporal dynamics of TF in solid tumor biopsies
To investigate the temporal dynamics of TF expression, IHC was per-  (Tables S2 and S3).
Tumor TF expression (H-score) was generally comparable between the first and second biopsy in patients with cervical, ovarian, and prostate cancer ( Figure 3A,B), although differences in H-score were observed between T1 and T2 for individual patients. In biopsies from patients with endometrial cancer, H-scores appeared to be lower in T2 than in T1 samples. However, sample size (n = 8) was not sufficiently large to draw firm conclusions to whether the differences in TF expression indeed reflect changes due to time or treatment, or reflect tumor heterogeneity. No trend for variation in TF expression over time was observed in patient-matched tumor biopsies of the remaining cancer types that were grouped together because they had insufficient numbers for statistical analysis (i.e., gastro-esophageal, lung, and bladder cancer) ( Figure 3C).
Variation in TF H-scores between T1 and T2 was not dependent on the time interval between the biopsies ( Figure 3D) and appeared to be unrelated to disease progression in prostate, cervical, ovarian, and endometrial cancer patients as depicted by early (E; stage 1 and 2) and advanced (A; stage 3 and 4) stages of disease ( Figure S4). Similarly, treatment received between T1 and T2, including chemotherapy, radiotherapy, or hormone treatment (Table S3), did not have a significant effect on TF expression at the population level (data not shown).

Although TF expression was not significantly different between T1
and T2 samples per cancer type and appeared to be independent of disease progression or treatment, differences were observed for some patients.

| DISCUSSION
TF is aberrantly expressed in a broad range of solid tumors compared to normal tissue, 8 and has been associated with poor prognosis. 12,13,18 Here we determined the prevalence, cellular localization, and the spatial and temporal expression patterns of TF in a broad panel of solid tumor biopsies taken at various time points and stages of disease, and from primary and metastatic lesions, using validated reagents and IHC methods.
The results in this manuscript indicate that TF expression in tumor cells is generally stable over time, is independent of disease progression or treatment regimen, and independent of tumor dissemination. These results may appear counter-intuitive as the literature suggests that TF expression may be increased with more advanced stages of disease, as a prothrombotic state typically characterizes more advanced malignancies 19 and is a significant cause of patient death. 20   spatial and temporal expression of TF in solid tumors. Normal tissues were not scored, because TF typically has low expression in most tissues as described previously. 5,6 We show that TF is expressed in tumor cells across a broad range of solid cancer types, with differences in TF expression prevalence and staining intensity between various solid tumor specimens. Distinctive patterns were observed across solid tumors, with pancreatic cancer, cervical cancer and HNSCC standing out with, predominantly membranous, TF expression in at least 75% of the cases. Interestingly we observed low prevalence of TF expression in breast, ovarian, and bladder cancer which are in discordance with previous reports. [23][24][25] This discordance may be a result of differences in methodology (e.g., detection antibodies or assay parameters), the selected patient population, the composition of the scored tissue samples, biological differences, or even publication bias, since negative results are less likely to be published. The present study restricted analysis of TF expression to tumor cells and did not take into account expression on other cells in the tumor microenvironment (e.g., stromal cells), which has been previously described in some solid tumors. 26 TF H-scores in primary tumors showed a significant correlation with patient-matched metastasis. At the population level, TF H-scores were generally stable over time irrespective of disease progression or treatment, although differences in H-scores between biopsies taken at different timepoints were observed for individual patients.
The broad expression across solid cancers makes TF a relevant therapeutic target and supports further investigation of TF-targeting agents across multiple tumor subtypes and various stages of disease.
These data suggest that TF might be considered as a biomarker for various solid tumors, but future large-scale studies are warranted.