Repurposing of a Thromboxane Receptor Inhibitor Based on a Novel Role in Metastasis Identified by Phenome-Wide Association Study

This article is featured in Highlights of This Issue, p. 2407

The use of deidentified human genetic data tied to robust electronic health records (EHR) enables discovery of human genotypes correlating with specific clinical conditions. This approach, known as phenome-wide association study (PheWAS; refs. 1, 2), can discover clinical “phenotypic” indications associating with a genotypic variation, enabling discoveries of novel pathophysiologic gene functions. From a translational perspective, PheWAS could predict potential opportunities for repurposing an existing catalog of molecularly targeted therapies (3), diminishing the risk, cost, and time needed for bringing treatments to patients.

To test this principle, we used PheWAS to investigate repurposing opportunities for the drug CPI211 (4–6), a potent and highly selective small-molecule inhibitor of thromboxane A2 and prostanoid receptor (TPr). CPI211 (PubChem ID 3037233) is safe and well tolerated, as shown in a randomized, dose-escalating (≤1,000 mg), placebo-controlled clinical trial (7). Decades of basic and translational science determined that TPr, expressed in platelets, endothelial cells (EC), and smooth muscle cells, regulates different aspects of cardiovascular homeostasis. TPr dysregulation contributes to venous hypertension, pulmonary hypertension, thrombosis, and asthma, steering the development of CPI211 and other TPr antagonists toward these clinical indications (8, 9). However, aberrant activation of TPr may have yet undiscovered pathologic consequences for which TPr inhibitors might be effective. Because pathologically increased TPr expression can be caused by naturally occurring single-nucleotide polymorphisms (SNP) within the TPr gene, TBXA2R, we propose that PheWAS analysis using these previously identified and characterized TBXA2R SNPs as a genomic proxy for increased TPr expression could identify clinical manifestations correlating with increased TPr expression (10).

Several previously described SNPs within TBXA2R are known to alter TPr expression, some that diminish TPr and thus correlate with decreased platelet function (11), and others that increase TPr expression. As might be expected, the SNPs that increase TPr expression are known to correlate with advanced cardiovascular disease (12–14) or asthma (15). We investigated one TBXA2R SNP that increases TPr expression, rs200445019. This SNP was selected based on its occurrence within the TBXA2R coding region, generating a threonine-to-alanine alteration at residue 399 (T399A) within the C-terminus of the TPr splice variant, TPr-β (https://www.ncbi.nlm.nih.gov/snp/rs200445019/), the domain of TPr-β driving TPr desensitization, internalization, trafficking, and proteosomal degradation (16–19). Following ligand activation, TPr signaling activates protein kinase C (PKC), a serine threonine kinase that then phosphorylates TPr-β T399, inducing TPr internalization and degradation. TPr T399 mutation causes loss of PKC-mediated TPr phosphorylation, and loss of ligand-induced receptor downregulation. Thus, the TPr-β T399A substitution generated by rs200445019 results in greater TPr expression and signaling, supporting use of rs200445019 as a genomic proxy for increased TPr in PheWAS analyses. Here, we report that PheWAS analysis of rs200445019 identified a novel and unexpected correlation between rs200445019 and metastatic cancer diagnoses, and that the TPr inhibitor CPI211 blocked metastasis in multiple cancer models.

The BioVU biorepository at Vanderbilt University Medical Center (VUMC) contains >250,000 deidentified DNA samples extracted from excess patient blood samples collected during routine clinical testing, linked to corresponding, longitudinal clinical and demographic data derived from Synthetic Derivative, a deidentified EHR built for research purposes (20, 21). BioVU data were utilized in accordance with VUMC IRB# 151121. The allelic variant of focus was the TBXA2R missense SNP rs200445019, selected based on the presence of meaningful validation signals in the PheWAS results to support inference of variant effects in vivo, including hypertension and cardiovascular indications, phenotypes expected to be observed if TPr function was enhanced.

PheWAS using previously reported methods (22) focused on 29,722 patients of European ancestry genotyped using the Illumina Infinium Exomechip (Illumina, Inc.), containing ∼250,000 SNPs across genomic protein coding regions discovered through exome- and whole-genome sequencing in more than 12,000 individuals. All distinct phenotypic data (ICD9 billing codes representing approximately 1,660 phenotypes) were captured from EHRs and translated into corresponding phenotype groupings. Cases for each phenotype (condition) were defined as patients having ≥ 2 ICD9 codes documented on separate days and mapping to a specific PheCode. The PheCode ‘malignant secondary neoplasm’ (PheCode 198) included ICD9 codes for secondary malignant neoplasm of respiratory and digestive systems, secondary malignant neoplasms of other specified sites, malignant pleural effusion, and several others. Controls were defined as all patients who did not have any of the ICD9 codes defining a case. Only phenotypes occurring in ≥25 cases (0.42% of genotyped patients) were included. Physicians at VUMC with relevant qualifications and content-specific knowledge performed evidence reviews in cases of secondary malignancy. The PheWAS algorithm (22) was applied to calculate case and control genotype distribution, χ² distribution, associated allelic P value and allelic odds ratio (OR). For χ² distribution cell counts below five, Fisher exact test was used to calculate P value using R software (http://www.r-project.org/).

CPI211 (Supplementary Fig. S1A, PubChem ID 3037233) was generously provided by Cumberland Pharmaceuticals, Inc. The chemical formula of CPI211 is C₂₅H₃₂N₂O₅, with a molecular weight of 440.5 g/mol. CPI211 is also known by the following aliases: Ifetroban, 143443-90-7, UNII-E833KT807K, and BMS-180291.

Cells purchased from American Tissue Type Collection [4T1: (CRL-2539) HEK-293T/17 (CRL-11268), MDA-MB-231 (HTB-26), MiaPaCa-2 (CRL-1420), A549 (CRM-CCL-185)] were maintained in growth medium defined by supplier. Cells constitutively expressing luciferase were generated by lentiviral particle (LVP) transduction with CAG-Luciferase LVP (GenTarget Inc.) followed by puromycin selection. LVP encoding human TPr-β and TPr-β(T399A) were custom synthesized in Lentifect (GeneCopoeia). 293T cells were transduced with empty LVP or those expressing TPrβ(WT) and TPr-β(T399A), followed by blasticidin selection. Mouse pulmonary microvascular ECs (MPMEC) were harvested and cultured as described (23). For MPMEC growth assays, cells (0.5 × 10⁴) were seeded on gelatin-coated 35-mm dishes, and cultured for 7 days in growth media, U46619 (0.5 μmol/L), CPI211 (0.1 μmol/L), or 4T1 conditioned media (serum-free media conditioned for 48-hour culture of 4T1 cells, passed through a 0.2-μm syringe filter). Cells were stained with 0.01% crystal violet, rinsed with water and dried, and then imaged on a flatbed scanner. Human umbilical vein ECs (HUVEC) were purchased from Lonza and maintained in EGM-2 medium (Lonza). All cells used were cultured for <15 passages.

Human platelets were collected from blood donated by healthy anonymous volunteers, and prepared as previously described (24). The platelet pellet was suspended in Tyrode's buffer containing 0.1% bovine serum albumin and counted on a Beckman Z1 Coulter counter.

Cells were homogenized in lysis buffer [50 mmol/L Tris pH 7.4, 100 mmol/L NaF, 400 mmol/L NaCl, 0.5% NP-40, 100 μmol/L Na₃VO₄, 1X protease inhibitor cocktail (Roche)] and cleared by centrifugation. Platelets were homogenized with 0.6 N perchloric acid and cleared by centrifugation. For immunoprecipitation, cell lysates (1000 μg) cleared with protein A/G+ agarose were incubated with anti-VE-cadherin antibody (5 μg) and protein A/G+ agarose. Samples were resolved on 4%–12% polyacrylamide gels (Novex), transferred to nitrocellulose membranes (iBlot), and Western analysis was performed as described (25) using the following primary antibodies: [α-actin (Sigma-Aldrich); TPr (Abcam); from Cell Signaling Technologies: protein kinase C phospho-substrate, phospho-myosin light chain (MLC), MLC, Vascular Endothelial (VE)-Cadherin, phospho-tyrosine horseradish peroxidase (PTyr-100)].

All animals were housed under pathogen-free conditions. Experiments were performed in accordance with AAALAC guidelines and Vanderbilt University Institutional Animal Care and Use Committee approval. Where indicated, CPI211 was delivered daily at 50 mg/kg by oral gavage in 25 μL vehicle (4% sucrose in sterile water). 1 × 10⁶ tumor cells (4T1, MDA-MB-231, A549, and MiaPaCa2) were injected via tail vein of athymic (nu/nu) Balb/C female mice. 4T1 tumor cells (1 × 10⁶) were injected into the left inguinal mammary fat pad of athymic (nu/nu) Balb/C or WT Balb/C female mice. Tumor dimensions were measured using calipers. Tumor volume was calculated as: volume = length × width² × 0.52). For bioluminescence imaging, mice were injected intraperitoneally with D-luciferin monosodium substrate (Thermo Fisher, 150 mg/kg) and imaged on IVIS Lumina III (Xenogen Corporation) 30 minutes after injection.

Tissue processing, staining, and immunohistochemistry (IHC) were performed by the Vanderbilt Translational Pathology Shared Resource. IHC on 5-μm paraffin-embedded sections was performed as described previously (26) using anti-CD31 (Abcam) and anti-Ki67 (Santa Cruz Biotechnologies). Photomicrographs were acquired on an Olympus CK40 inverted microscope through an Optronics DEI-750C camera using CellSens software for capture and morphometric analyses (27).

MPMECs or HUVECs (1 × 10⁵) were seeded on Matrigel-coated transwell lower surfaces and cultured 1 week before labeling with CellTracker Red (Molecular Probes; 0.5 mg/mL), transferred to 24-well dishes in 2% serum ± U46619 (0.5 μmol/L) and/or CPI211 (0.1 μmol/L) for 6 hours. 4T1-GFP cells (2 × 10⁵) were seeded into the upper chamber for 16 hours, swabbed from upper transwell surfaces and then visualized on the lower filter surface using fluorescence microscopy.

Treatment groups were compared using either two-tailed Student t test or one-way ANOVA test coupled with Tukey means comparison test, where a P value < 0.05 was deemed representative of a significant difference between treatment groups.

The TPr-T399A alteration generated by rs200445019 reportedly impairs TPr desensitization and degradation following ligand-induced stimulation. To confirm this, we expressed wild-type TPr-β (TPrβ-WT) and TPr-β T399A (TPrβ-T399A) in 293T cells, which express low endogenous TPr levels (Supplementary Fig. S1B). Under serum-starved conditions, similar protein expression of TPr was seen in cells expressing TPrβ-WT and TPrβ-T399A. As expected, addition of the TPr agonist U46619 (0.5 μmol/L) increased PKC activity and MLC phosphorylation (Fig. 1A), which occurred at similar levels in cells expressing TPrβ-WT and TPrβ-T399A. However, TPrβ-WT protein levels diminished within 5 minutes of U46619 treatment, whereas TPrβ-T399A levels remained unchanged, consistent with previous reports that TPrβ T399 drives ligand-induced TPr downregulation. This was confirmed by treating cells with U46619 for 5 minutes, then chasing with U46619-free media for up to 60 minutes, revealing a sustained downregulation of TPrβ-WT, but persistently elevated TPrβ-T399A (Fig. 1B). These findings confirm use of rs200445019 as a proxy for increased TPr expression.

To detect phenotypes associated with rs200445019, we leveraged the BioVU database (21) to query 29,722 deidentified patient EHRs and their associated genotyping, scanning for novel associations between PheWAS codes and rs200445019. As expected, our unbiased PheWAS approach identified associations of rs200445019 with chronic venous hypertension, pulmonary heart disease, and primary pulmonary hypertension (Table 1; ref. 14). Unexpectedly, PheWAS also identified a novel association between rs200445019 and metastatic disease (coded as Secondary Malignancy) at multiple tissue sites, including lymph nodes, respiratory organs, digestive systems, brain/spine, and others. Chart review by medical experts confirmed the diagnoses of secondary malignancies, and revealed that these metastatic lesions were derived from multiple solid tumor types, including breast, colon, lung, head and neck, renal, gastric, ovarian cancers, melanomas, and others (Fig. 1C). However, PheWAS did not detect a specific correlation of rs200445019 with any primary cancer diagnosis, per se, suggesting that increased TPr expression and/or activity might have a selective role in tumor metastasis.

Next, we assessed the relationship between TBXA2R expression and cancer patient outcome. Analysis of METABRIC-curated invasive breast cancers assessed by gene-expression array (N = 2,051; ref. 28) revealed decreased overall survival (OS) of patients whose tumors harbored high TBXA2R (Fig. 1D). High TBXA2R expression also correlated with decreased OS in lung adenocarcinomas and gastric cancers (29). We extended this analysis across 14 additional cancer types, using publicly available tumor RNA-seq data tied to disease-free progression (DFP) outcomes (ref. 29; Supplementary Table S1), finding that 12 of 15 solid cancer types displayed an inverse correlation between TBXA2R expression and DFP, with strong statistical significance in 11 of these, including breast cancer, ovarian cancer, renal papillary cell carcinoma, sarcoma, and others. Interestingly, 2 of 14 cancer types assessed showed increased DFP in patients expressing increased TBXA2R [bladder cancer and hepatocellular carcinoma (HCC)]. The reasons underlying why bladder cancer and HCC might show improved survival upon increased TBXA2R expression are unclear, but may relate to unique pathophysiologies of these cancers, routes of metastatic spread, or tissue-selective roles of TPr. For the 12 cancers demonstrating a correlation between TBXA2R expression and worse outcome, these findings support the notion that TBXA2R/TPr may have an underappreciated role in tumor metastasis, and that targeted TPr inhibition might suppress metastasis.

To confirm TPr inhibition by the TPr antagonist CPI211, we assessed 293T cells expressing exogenous TPr-β. Western blot analysis confirmed that CPI211 blocked MLC phosphorylation in U46619-treated cells expressing TPrβ-WT and TPrβ-T399A (Fig. 1E). We also confirmed TPr-α inhibition by CPI211 using washed human platelets, which almost exclusively express TPr-α. U46619-induced P-MLC was blocked robustly by CPI211 (Fig. 1F). Dilutions of CPI211 down to 5 nmol/L blocked U46619-induced platelet aggregation (Fig. 1G), confirming potent TPrα inhibition by CPI211.

We assessed the impact of TPr inhibition on hematogenous metastasis using tumor cells [4T1 (mouse mammary cancer), MDA-MB-231 (human breast cancer), MiaPaCa2 (human pancreatic cancer), and A549 (human lung cancer)] delivered directly into circulation by intravenous (i.v.) injection. Western blot analysis revealed only very low TPr expression in each tumor cell line, and modest or no P-MLC in U46619-treated cells (Fig. 2A). Athymic (nu/nu) mice were pretreated with CPI211 (50 mg/kg) before tumor cell delivery, continuing treatment for the following 28 days (Fig. 2B). 4T1 lung metastases were seen in 100% of vehicle-treated mice, but only 70% of CPI211-treated mice (N = 10; Fig. 2C). Similarly, CPI211 treatment decreased the percentage of mice harboring MDA-MB-231 lung metastases from 90% to 20% (N = 10), and mice with A549 lung metastases from 60% to 10% (N = 10). Although 90% of mice inoculated with MiaPaca2 cells developed lung lesions regardless of vehicle or CPI211 treatment, the average number of MiaPaca2 metastases per mouse was diminished in CPI211-treated mice by more than half. These results suggest that TPr inhibition decreases hematogenous metastasis of multiple cancer types.

Because venous tumor cell delivery allows tumor cells to bypass steps of intravasation, we used a rigorous model of spontaneous metastasis from primary orthotopic 4T1 mouse mammary tumors. Mice harboring luciferase-expressing 4T1 tumors were randomized into treatment groups receiving vehicle or CPI211 when tumors reached 50 mm³ (Fig. 2D). Mice were treated daily for 28 days, although tumors were surgically resected on treatment day 12. By treatment day 21, luminescence was detected in the thoracic region of vehicle-treated mice, but not in CPI211-treated mice (Fig. 2E). By day 28, average luminescence in vehicle-treated mice was approximately 18-fold higher than in CPI211-treated mice. Notably, 3 of 10 mice treated with CPI211 had no measurable luminescence above baseline (Fig. 2E, right). Histologic examination of lungs confirmed reduced lung metastases in CPI211-treated mice versus vehicle-treated mice, including an absence of metastases in 3 of 10 CPI211-treated mice.

We measured volume of tumors grown in athymic mice and treated for 12 days with CPI211 beginning when tumors were 50 mm³, revealing no difference in tumor size between vehicle and CPI211-treated groups (Fig. 3A). IHC for the cellular proliferation marker Ki67 revealed similar Ki67 staining in tumors harvested from vehicle-treated and CPI211-treated mice (Fig. 3B and C). These findings were supported by measurements of tumor cell growth in culture, showing that neither TPr activation nor TPr inhibition affected growth of 4T1 or MDA-MB-231 cells (Fig. 3D).

An abundance of large vascular structures seen in vehicle-treated tumors was substantially diminished in CPI211-treated tumors, as shown by staining with the vessel endothelial marker CD31 (Fig. 3E). Software-based morphometric analyses of CD31-stained tumor sections revealed a decreased total area of CD31⁺ vessels in CPI211-treated tumors (Fig. 3F) and fewer CD31⁺ structures exceeding 50 μm² (Fig. 3G). Thus, it is possible that TPr signaling affects the tumor vasculature in a manner that promotes metastasis.

We investigated the impact of TPr signaling on ECs using primary cultures of mouse pulmonary microvascular ECs (MPMEC), which abundantly expressed TPr (Fig. 4A). U46619 induced PKC activity in MPMECs. As expected, TPr expression was not diminished in U46619-treated MPMECs, because mouse cells express only TPr-α, which, unlike TPr-β, is not internalized in response to ligand activation (19). Importantly, PKC substrate phosphorylation was potently inhibited by CPI211. Similar PKC substrate phosphorylation was seen in Western blot analyses of primary HUVECs treated with U46619, which was blocked upon treatment with CPI211. Notably, TPr expression diminished in U46619-treated HUVECs, which abundantly express TPr-β (30). Agonist-induced TPr diminution was blocked by CPI211 in HUVECs.

Growth of MPMECs seeded at low density over 7 days was unaffected by U46619 and/or CPI211 (Fig. 4B). Interestingly, serum-free 4T1 tumor cell–conditioned media increased MPMEC growth, although this, too, was unaffected by CPI211.

Accumulating evidence shows that EC barrier function becomes compromised in cancers (31–33). We measured tumor cell migration across an endothelial barrier using a modified transwell approach (Fig. 4C). CellTracker Red-labeled MPMECs seeded on Matrigel on the lower side of a transwell filter were grown into a barrier monolayer, and then treated with U46619 prior to adding 4T1-GFP cells into the upper chamber, assessing transendothelial migrations by fluorescence microscopy. These studies revealed increased 4T1 transendothelial migration across U46619-treated MPMECs as compared with vehicle-treated MPMECs (Fig. 4D and E). Similarly, an increased number of MDA-MB-231 cells migrated across U46619-treated HUVECs as compared with vehicle-treated HUVECs. In both cases, CPI211 blocked tumor cell migration across the EC barrier. However, in the absence of ECs, 4T1 cell migration across Matrigel-coated transwells was unaffected by U46619 or by CPI211 (Supplementary Fig. S2A), as was MDA-MB-231 migration (Supplementary Fig. S2B). These findings suggest that TPr signaling may not affect tumor cell migration, per se, but may enhance transendothelial migration of tumor cells.

Next, we measured the barrier function of MPMEC monolayers cultured on transwells and treated with U46619 and/or CPI211. Transwell filters were transferred to fresh wells containing only PBS after 6 hours, at which point albumin (50 μg/mL) was added to the upper transwell chamber. After 2 minutes, the lower chamber was sampled for albumin distribution (Fig. 4F). U46619-treated MPMECs showed increased albumin passage to the lower chamber, which was blocked by CPI211. These results suggest that TPr signaling may increase permeability of an endothelial barrier, and are consistent with previous reports suggesting that thromboxane signaling may decrease vascular barrier function (34, 35).

VE-cadherin localization at EC junctions is critical for endothelial barrier function. Notably, VE-cadherin tyrosine phosphorylation induces VE-cadherin internalization, compromising endothelial barrier function (36), enabling metastatic dissemination of tumor cells (37, 38). Using immunofluorescence (IF) to assess VE-cadherin localization in MPMECs cultured on transwell filters, we identified punctate patterns of VE-cadherin staining along cell membranes at or near cell–cell contacts in vehicle-treated MPMECs (Fig. 4G), whereas U46619-treated MPMECs displayed diffuse intracellular VE-cadherin staining. Although U46619 did not affect total VE-cadherin expression (Fig. 4H), U46619-induced VE-cadherin tyrosine phosphorylation, which was blocked by CPI211 (Fig. 4H), suggesting that TPr signaling may induce VE-cadherin tyrosine phosphorylation and internalization, leading to decreased barrier function.

Next, we established models of spontaneous metastasis from orthotopic 4T1 tumors grown in immune-competent isogenic Balb/C mice, with two separate CPI211 treatment schemes assessed. The first approach modeled neoadjuvant CPI211 treatment (i.e., without tumor resection), in which tumor-bearing mice were randomized into treatment groups when tumors reached 50 mm³. Mice were treated daily for 28 days with vehicle or CPI211 (Fig. 5A). Although lung metastases were seen in all vehicle-treated mice (N = 10), only 60% of CPI211-treated mice harbored lung metastases (Fig. 5B). Of the mice that developed lung metastases, fewer lung metastases were seen CPI211-treated samples versus vehicle, demonstrating that TPr inhibition diminishes spontaneous metastases in the neoadjuvant setting.

We next tested a model of adjuvant TPr inhibition following surgical resection of 300 mm³ tumors. Mice were randomized for treatment 24 hours after tumor resection (Fig. 5C), receiving daily treatment with vehicle or CPI211 for 21 days. Although 100% of vehicle-treated mice and 90% of CPI211-treated mice developed lung metastases (Fig. 5D), the number of lung metastases per mouse was reduced in CPI211-treated mice to half of what was seen in controls. These studies reveal a potential benefit for reducing metastatic burden through TPr inhibition, even after tumor cells may have entered circulation.

We identified a novel and unexpected correlation between TPr and tumor metastasis using a combination of PheWAS analyses in large population data sets, gene-expression studies in clinical cancer data sets, and pharmacologic blockade of TPr activity in models of metastasis (Supplementary Fig. S3). Importantly, the TPr inhibitor CPI211 decreased both experimental (hematogenous) and spontaneous tumor metastasis.

The metastatic process requires tumor cells to overcome a series of physiologic barriers (2). Tumor cells must invade through primary tumor extracellular matrix to reach the vasculature, migrate across the vascular endothelium into circulation, survive in circulation, evade immune recognition, and adhere to and migrate across the endothelium into distant tissue, where tumor cells must survive and grow (39). Findings shown herein highlight an underappreciated role for TPr signaling in disrupting the endothelial barrier in the metastatic process, facilitating tumor metastasis. These studies build upon published studies demonstrating that platelets, a rich source of the TPr ligand TxA2, are potent mediators of metastasis through a range of pathways involving interactions between tumor cells, ECs, and platelets (40–42). Platelet-CTC aggregation may enhance metastasis by providing important survival signals to the tumor cell, by shielding CTCs from immune surveillance, and by promoting adhesion to the vascular endothelium. In fact, previous studies report that TBXA2R-null mice display reduced metastases in experimental (tail-vein) models of metastasis (42). Thus, CPI211 may reduce platelet-assisted metastatic processes. Although we have not tested this hypothesis specifically, we confirmed that CPI211 inhibits platelet aggregation (Fig. 1).

Although platelet TPr likely has a prominent role in facilitating tumor metastasis, we show that endothelial TPr signaling increases transendothelial tumor cell migration, even in the absence of platelets (Fig. 4). Endothelial TPr signaling diminished VE-cadherin at endothelial cell–cell junctions. Interestingly, published studies show that hypoxia contributes to VE-cadherin endocytosis in ECs, resulting in endothelial barrier disruption and vascular dysfunction, while increasing TPr expression and sensitivity (38, 43), together suggesting that tumor hypoxia might increase endothelial TPr signaling, thus facilitating metastasis, although this hypothesis remains to be tested.

Recent studies highlighted a role for COX-1 in generation of a premetastatic niche within the vasculature of colorectal cancers, in large part through its ability to increase TxA2 production, thus enhancing platelet–tumor cell aggregation upon vessel walls. Inhibition of COX-1 using aspirin decreased metastasis through downregulation of platelet TxA2 (41), implicating TPr signaling in metastasis. These studies hold great clinical significance, as the Add-Aspirin study (NCT02804815) is currently testing if daily aspirin use (100 mg, 300 mg, or placebo, for 5 years) after SOC treatment prevents recurrence and improves survival in patients lacking clinically evident metastases at the conclusion of SOC treatment.

It is important to note that COX-1 produces both TxA2 and gut-protecting prostaglandin E2 (PGE2). As such, aspirin may have dangerous consequences for some patients. It is exciting to speculate TPr inhibition could provide the antimetastatic benefits of COX1 pathway inhibition, but without the potentially harmful side effects in aspirin-sensitive patients. However, given the potent antagonism of platelet coagulation by CPI211, studies going forward should proceed with awareness of bleeding or other potential adverse events that may arise in CPI211-treated patients.

Because TPr inhibition diminished metastasis without impacting tumor growth, it is anticipated that TPr inhibition could be effective in combination with SOC cytotoxic therapies in the adjuvant or neoadjuvant setting. Data presented herein suggest that TPr inhibitors could provide greatest benefit as a preventive treatment, a clinical course of action not frequently used. However, this approach is being tested in at least one clinical trial (NCT03694249), in which patients diagnosed with aggressive cancers with a high risk of recurrence (e.g., gastric cancer; triple-negative breast cancer and others) but lacking clinically identifiable metastases will be randomized post-surgically into treatment groups receiving CPI211 or placebo for 12 months, and followed for an additional 12 months. This landmark clinical trial will test the feasibility of a pharmaco-preventive strategy that may diminish metastatic burden in certain cancer patients.

The notion that TPr inhibitors will not eliminate existing metastases must be considered. However, TPr inhibition may reduce spread of tumor cells from secondary to tertiary tumor sites. Timing TPr inhibitor administration requires additional study, as does testing of other TPr inhibitors in clinical development (terutroban, ridogrel, terbogrel, and others; ref. 8). Further, it will be essential to identify which cancer types are responsive to TPr inhibition, necessitating a greater understanding of how different cancers use different metastatic routes.

In summary, we introduce TPr as an antimetastatic target across many types of cancers. Findings herein support the continued investigation of agents interfering with steps of the metastatic cascade, and support the use of PheWAS analysis to accelerate the pace of drug repurposing.

CPI211 used in this study was provided by Cumberland Pharmaceuticals, Inc. (CPI). CPI had no role in study design, experimental conduct, data collection and analysis, or preparation of the manuscript. Additionally, CPI provided no direct financial support for this work.

T.A. Werfel: Investigation, writing–original draft. D.J. Hicks: Investigation. B. Rahman: Investigation. W.E. Bindeman: Investigation. M.T. Duvernay: Investigation. J.G. Maeng: Data curation, investigation. H. Hamm: Resources, supervision. R.R. Lavieri: Conceptualization, data curation, investigation, writing–review and editing. M.M. Joly: Conceptualization, investigation, writing–review and editing. J.M. Pulley: Conceptualization, resources, funding acquisition. D.L. Elion: Investigation, writing–review and editing. D.M. Brantley-Sieders: Validation, investigation, writing–review and editing. R.S. Cook: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, writing–original draft, project administration, writing–review and editing.

We kindly thank Cumberland Pharmaceuticals, Inc., and Dr. Ines Macias-Perez for providing CPI211. We acknowledge Drs. Ingrid Meier and Justin Balko for thoughtful discussion. We acknowledge Vanderbilt Shared Resources: the VICC Breast SPORE Pathology (Dr. Melinda Sanders), Translational Pathology, Digital Histology, and VANTAGE shared resources. This work was supported by grants NIH P50 CA098131, NIH P30 CA68485, National Center for Advancing Translational Sciences CTSA UL1TR000445, and Congressionally Directed Medical Research Program W81XWH-161-0063.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.