Article Text

Original research
Blocking VCAM-1 inhibits pancreatic tumour progression and cancer-associated thrombosis/thromboembolism
  1. Makoto Sano1,2,
  2. Ryota Takahashi1,
  3. Hideaki Ijichi1,3,
  4. Kazunaga Ishigaki1,
  5. Tomoharu Yamada1,
  6. Koji Miyabayashi1,
  7. Gen Kimura1,
  8. Suguru Mizuno1,
  9. Hiroyuki Kato1,
  10. Hiroaki Fujiwara1,
  11. Takuma Nakatsuka1,
  12. Yasuo Tanaka1,
  13. Jinsuk Kim2,
  14. Yohei Masugi4,
  15. Yasuyuki Morishita5,
  16. Mariko Tanaka5,
  17. Tetsuo Ushiku5,
  18. Yousuke Nakai1,6,
  19. Keisuke Tateishi1,
  20. Yukimoto Ishii2,
  21. Hiroyuki Isayama7,
  22. Harold L Moses8,
  23. Kazuhiko Koike1
  1. 1 Department of Gastroenterology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
  2. 2 Division of Medical Research Planning and Development, Nihon University School of Medicine, Tokyo, Japan
  3. 3 Clinical Nutrition Center, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
  4. 4 Department of Pathology, Keio University School of Medicine, Tokyo, Japan
  5. 5 Department of Pathology, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
  6. 6 Department of Endoscopy and Endoscopic Surgery, Graduate School of Medicine, The University of Tokyo, Tokyo, Japan
  7. 7 Department of Gastoroenterology, Juntendo University School of Medicine, Tokyo, Japan
  8. 8 Department of Cancer Biology, Vanderbilt-Ingram Cancer Center, Nashville, Tennessee, USA
  1. Correspondence to Dr Hideaki Ijichi, Department of Gastroenterology, The University of Tokyo Graduate School of Medicine Faculty of Medicine, Tokyo 113-8655, Japan; hideijichi-gi{at}


Objective Pancreatic ductal adenocarcinoma (PDAC) is the deadliest cancer. Cancer-associated thrombosis/thromboembolism (CAT), frequently observed in PDAC, is known as a poor prognostic factor. Here, we investigated the underlying mechanisms between PDAC and CAT, and performed a trial of therapeutic approach for PDAC using a genetically engineered mouse model, PKF (Ptf1acre/+;LSL-KrasG12D/+;Tgfbr2flox/flox ).

Design Presence of CAT in PKF mice was detected by systemic autopsy. Plasma cytokines were screened by cytokine antibody array. Murine and human plasma atrial natriuretic peptide (ANP) and soluble vascular cell adhesion molecule 1 (sVCAM-1) were determined by ELISA. Distribution of VCAM-1 in PKF mice and human autopsy samples was detected by immunohistochemistry. PKF mice were treated with anti-VCAM-1 antibody and the effects on survival, distribution of CAT and the tumour histology were analysed.

Results We found spontaneous CAT with cardiomegaly in 68.4% PKF mice. Increase of plasma ANP and sVCAM-1 was observed in PKF mice and PDAC patients with CAT. VCAM-1 was detected in the activated endothelium and thrombi. Administration of anti-VCAM-1 antibody to PKF mice inhibited tumour growth, neutrophil/macrophage infiltration, tumour angiogenesis and progression of CAT; moreover, it dramatically extended survival (from 61 to 253 days, p<0.01).

Conclusion Blocking VCAM-1/sVCAM-1 might be a potent therapeutic approach for PDAC as well as CAT, which can contribute to the prognosis. Increase of plasma ANP and sVCAM-1 might be a diagnostic approach for CAT in PDAC.

  • pancreatic cancer
  • cell adhesion molecules

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. All data relevant to this study are included in the article and supplementary data.If you need to contact us, email address

Statistics from

Significance of this study

What is already known on this subject?

  • Xenograft models have been used for analysis of cancer-associated thrombosis/thromboembolism (CAT).

  • CAT is not detected in a typical pancreatic ductal adenocarcinoma (PDAC) mouse model KPC (Pdx1-cre;LSL- KrasG12D/+;LSL-Tp53R172H/+ ).

  • Increase of soluble vascular cell adhesion molecule 1 (sVCAM-1) is reported in non-cancer patients with thrombosis/thromboembolism.

  • VCAM-1 is highly expressed in cancer tissues including PDAC and is associated with cancer progression.

What are the new findings?

  • Spontaneous CAT with cardiomegaly was frequently observed in a PDAC mouse model PKF (Ptf1acre/+;LSL-KrasG12D/+;Tgfbr2flox/flox ).

  • Increase of plasma atrial natriuretic peptide (ANP) and sVCAM-1 was detected in PKF mice and PDAC patients with CAT.

  • Blocking VCAM-1/sVCAM-1 in PKF mice delayed tumorigenesis by inhibiting local infiltration of neutrophils and macrophages, inhibited progression of CAT and extended the survival more than four times longer than the control.

How might it impact on clinical practice in the foreseeable future?

  • The strong survival effect of VCAM-1 blockade might be a potent therapeutic method for PDAC which can improve the poorest prognosis.

  • Blocking VCAM-1/sVCAM-1 can be an inhibiting approach for CAT, thereby supporting better prognosis.

  • Plasma ANP and sVCAM-1 can be a diagnostic approach for CAT in PDAC.


Pancreatic cancer is the third and fourth leading cause of cancer death in the world.1 2 Five-year survival rate is still 8%, indicating the worst prognosis in cancers. Pancreatic ductal adenocarcinoma (PDAC) is the most common type; therefore, effective therapeutic strategy that can improve the prognosis of PDAC is one of the priorities in unmet medical needs.

Cancer-associated thrombosis/thromboembolism (CAT) is a major complication and cause of death in cancers including PDAC.3–5 CAT is detected in more than 8% of overall patients with PDAC, classified as a high-risk group.6 7 Most patients with PDAC are given chemotherapy, which can usually benefit the patients’ prognosis8; however, chemotherapy can also increase incidence of CAT, which was significantly associated with poor prognosis of patients with PDAC in previous studies.5 7 Therefore, preventing CAT may improve the prognosis of patients with PDAC.

It is known that CAT in PDAC is associated with procoagulant factors, proinflammatory cytokines and mucins.6 PDAC-induced platelet aggregation and expression of various procoagulant factors such as tissue factor (TF) induce hypercoagulable state, while proinflammatory cytokines tumour necrosis factor-α (TNF-α) and interleukin-1 (IL-1) as well as overexpression of cyclo-oxygenase-2 (COX-2) activate the process of thrombosis.9 10 Aberrant expression and glycosylation of mucins are associated with CAT and Trousseau’s syndrome, CAT-associated thrombophlebitis.11 12

CAT is related to activation of oncogenes such as KRAS and inactivation of tumour suppressor genes TP53.13 14 Activation of KRAS is observed in more than 95% of PDAC cases, while inactivation of TP53 has been shown to increase in frequency with progression of murine pancreatic intraepithelial neoplasm (mPanIN) to PDAC.15 Interestingly, however, CAT is uncommon and has never been documented in a murine PDAC model containing mutant KrasG12D and Tp53R172H expression, KPC (Pdx1cre/+;LSL-KrasG12D/+;LSL-Tp53R172H/+ ).16 17 While we have reported a murine PDAC model, PKF (Ptf1acre/+;LSL-KrasG12D/+;Tgfbr2flox/flox ), harbouring pancreatic epithelium-specific expression of KrasG12D and knockout of TGF-β type II receptor (Tgfbr2), which recapitulates histopathological features of human PDAC and cancer–stromal interactions,18–20 systemic CAT in PKF mice has not been investigated.

Vascular cell adhesion molecule-1 (VCAM-1) is an intercellular adhesion molecule, which is expressed on the surface of activated endothelial cells in an inflammatory condition and facilitates adhesion of leucocytes to endothelial cells.21 VCAM-1 is known to be cleaved from cell surface and released into the blood stream in a soluble form of VCAM-1 (sVCAM-1).22 It has also been reported that VCAM-1 is expressed on many types of cancer cells, including pancreatic cancer.23–27 In breast cancer, VCAM-1 was suggested to be involved in metastasis formation.23 28 However, the role of VCAM-1/sVCAM-1 in pancreatic cancer has not been clear.

In this study, we found that PKF mice frequently developed systemic CAT, accompanied with cardiomegaly. Increased atrial natriuretic peptide (ANP) and sVCAM-1 in the plasma were associated with CAT in both PKF mice and patients with PDAC. Moreover, administration of VCAM-1-neutralising antibody to the PKF mice delayed PDAC development by inhibiting immune-inflammatory cell infiltration and reduced systemic CAT, with a dramatically prolonged overall survival. Therefore, blockade of VCAM-1/sVCAM-1 can be a potential therapeutic strategy that contributes to the PDAC prognosis, and plasma ANP and sVCAM-1 can have a diagnostic value for CAT.

Materials and methods

Materials and methods are shown in the online supplemental data.

Supplemental material


CAT is systemically observed in PKF mice, accompanied with cardiomegaly and increased plasma ANP

First, we performed systemic autopsy of murine PDAC model PKF, when the mice got moribund, to determine the presence and distribution of CAT. We found organising thrombus in 13 of 19 (68.4%) PKF mice (online supplemental table 1), while no thrombus was found in wild-type (control) mice at 10 weeks of age. There was no significant difference in age of PKF mice with CAT and without CAT, 73 (54–144) days vs 74 (56–144) days (median and range of age), respectively (online supplemental table 1). In another PDAC model, LSL-KrasG12D/+;Trp53flox/+;Pdx-1cre/+ (KPCflox ),29 CAT was rarely observed (online supplemental table 2).

Interestingly, mural thrombi were detected in both ventricular walls, pulmonary arteries as well as portal vein in the liver in the PKF mice with CAT (figure 1A). In addition, occasional thrombi were detected in aorta and inferior vena cava (IVC) with marked dilation (online supplemental figure 1A,B). Histopathologically, almost all the thrombi were organising and were positive for Masson’s trichrome (MT) and phosphotungstic acid haematoxylin (PTAH) partially (figure 1A). The organising thrombi were positive for immunostaining by antiserum to von Willebrand factor (vWF) (figure 1B), while PTAH-negative fresh thrombi in small-sized blood vessels were also positive for vWF (figure 1B), indicating that platelets and fibrin were both involved in the thrombi.

Figure 1

Thrombosis/thromboembolism in a murine pancreatic cancer model PKF (Ptf1acre/+;LSL-KrasG12D/+;Tgfbr2flox/flox ). (A) Organising thrombi (asterisk) on cardiac ventricular walls (×20), pulmonary arteries and hepatic portal veins (×40) were stained by H&E, Elastica van Gieson (EVG), Masson’s trichrome (MT) and phosphotungstic acid haematoxylin (PTAH) staining. (B) Immunohistochemistry for von Willebrand factor (vWF) of organising thrombi in the heart (×20) and the liver (×40) as well as fresh thrombus in small-sized blood vessels in the pancreas (arrow; ×400). Right panel shows PTAH staining for fresh thrombus (arrow) in the pancreas (×400). Scale bars: 500 µm (×20), 200 µm (×40) and 100 µm (×400), respectively. (C) Weight (body weight ratio) of pancreas, heart and gastrocnemius muscle in control mice with normal pancreas (Cont, five male and five female), PKF mice without cancer-associated thrombosis/thromboembolism (CAT) (non-CAT, four male and two female) or with CAT (CAT, seven male and six female). (D) Plasma atrial natriuretic peptide (ANP) level in control mice with normal pancreas (Cont, four male and three female) and PKF mice without CAT (non-CAT, three male and three female) or with CAT (CAT, three male and four female). Mean value and SD are shown in the graphs. *p<0.05; **p<0.01.

The incidence of CAT was closely related to the weight of the pancreatic tumour in PKF mice (figure 1C). Weight of the heart and gastrocnemius muscle was significantly decreased in PKF without CAT, indicating cachexia, while weight of the heart was significantly increased in PKF with CAT, indicating cardiomegaly (figure 1C). Next, we investigated the plasma level of ANP, a marker of heart dysfunction, in PKF mice. As we expected, significantly higher level of ANP was detected in PKF mice with CAT compared with PKF without CAT as well as wild-type mice (figure 1D). PKF mice without CAT also showed a tendency of higher ANP levels compared with wild-type mice, but without statistical significance.

VCAM-1/sVCAM-1 is involved in CAT in PKF mice

To study the underlying mechanisms of frequent CAT in PKF mice, we compared the protein level of 144 cytokines in the plasma of PKF mice with or without CAT as well as wild-type mice by a cytokine array. sVCAM-1, ICAM-1, selectins and matrix metalloproteinases (MMPs), which are known to be risk factors for thrombosis,30 were elevated in PKF mice without CAT compared with wild-type mice (online supplemental figure 2). Additional risk factors for thrombosis, such as hepatocyte growth factor (HGF), thrombopoietin (TPO), leptin30 and a possible thrombosis marker CD3631 were also increased in PKF mice (online supplemental figure 2). Most of the other cytokines such as interleukins (ILs), TNF, proteases, chemokines, growth factors and stimulation factors were also elevated in the plasma of PKF mice than wild-type mice (online supplemental figures 2–11). These results indicate that PKF mice have hypercytokinemic condition and most of the cytokine levels in the plasma tended to decrease in the mice with CAT.

Among those secreted factors, we focused on sVCAM-1. ELISA revealed that PKF mice, with or without CAT, had significantly higher plasma level of sVCAM-1 compared with the control (figure 2A). There was no statistical difference in the level between the PKF mice with and without CAT. While VCAM-1 is known to be expressed on the surface of activated endothelial cells, immunohistochemical analysis in PKF mice revealed that VCAM-1 was highly expressed in the tumour cells and some of the fibroblasts (figure 2B), which indicated that the tumour cells were the major source of sVCAM-1 in the plasma. Notably, VCAM-1 was also detected in both fresh and organising thrombi in PKF mice, especially in fresh thrombi and in the central area of the organising thrombi (figure 2C), which indicated that VCAM-1 was highly involved in the thrombus formation. Interestingly, numerous leucocytes were agglutinated on the VCAM-1-positive endothelial cells (figure 2C–E). In fresh thrombotic lesion, CD11b+Ly-6G+ myeloid-derived suppressor cells (MDSCs) were observed around intima of endothelium (figure 2D). Meanwhile, a few myeloperoxidase (MPO)+ neutrophils and F4/80+ monocytes/macrophages were detected around intima of endothelial cells and K-19+ PDAC cell was not detected in the endothelium. In organising thrombus on IVC, Ly-6G+ MDSCs were accumulated on intima of IVC, and F4/80+ monocytes/macrophages could be surrounding the MDSC lesions (figure 2E). MPO+ neutrophils were mainly located in the centre of thrombus. These results indicated that a variety of immune-inflammatory cells and VCAM-1 were involved in the formation and progression of CAT.

Figure 2

Detection of vascular cell adhesion molecule 1 (VCAM-1) and leucocytes in cancer-associated thrombosis/thromboembolism (CAT). (A) Plasma levels of soluble VCAM-1 (sVCAM-1) in the pancreas of wild-type mice (Control, four male and four female), PKF mice without CAT (non-CAT, three male and three female) or with CAT (CAT, three male and four female). Mean value and SD are shown. **p<0.01. (B) Immunohistochemistry for VCAM-1 in the pancreatic ductal adenocarcinoma (PDAC) of PKF mice (×400). (C) Immunohistochemistry for VCAM-1 in endothelial cells (arrowheads; ×200) and aggregation of leucocytes (arrows, left), fresh thrombi (mid, arrows; ×100) and organising thrombus (right, asterisk, ×100) in the pancreas of PKF mice. (D) Localisation of leucocytes in fresh thrombus (asterisk) shown by H&E staining and immunohistochemistry for indicated molecules (×400). Dot lines, endothelial intima. Arrows indicate positively stained cells (CD11b+Ly-6G+, myeloperoxidase (MPO)+ and F4/80+ cells. Magnification: ×100 (left) and ×400 (mid) for H&E, and ×400 for immunohistochemistry. Insets are representative images of positive staining in fresh thrombus. Scale bars: 100 µm. (E) Localisation of leucocytes in organising thrombus (asterisk) shown by immunohistochemistry for indicated molecules (×400). Dot lines: endothelial intima. Scale bars: 100 µm.

Anti-VCAM-1 antibody dramatically extends PKF mice survival by regulating CAT and immune-inflammatory microenvironment

Next, we treated PKF mice with neutralising antibody to VCAM-1 or control immunoglobulin G (IgG), starting at 4 weeks of age and continued (n=5 in each group) (figure 3A). Kaplan-Meier analysis revealed that treatment with VCAM-1-neutralising antibody dramatically prolonged the survival of PKF mice compared with the control IgG injection (p<0.01 by log-rank test; figure 3B). Median survival time of the mice treated with anti-VCAM-1 antibody and control IgG were 253 and 61 days, respectively. We directly compared the pancreatic tumours using 3-week treatment with anti-VCAM-1 or control IgG (n=6 in each group). Notably, the weight of the pancreatic tumour was significantly reduced compared with the control (figure 3C), indicating an inhibitory effect of anti-VCAM-1 antibody on the PDAC progression. Histological analysis under H&E staining indicated that progression of mPanINs, pancreatic premalignant lesions and the sequential PDAC was delayed by administration of anti-VCAM-1 antibody (figure 3D,E), whereas Ki-67+ acinar-ductal metaplasia, one of the important sources of mPanIN, widely remained (figure 3D,F) and no significant difference was observed in terminal deoxynucleotidyl transferase -mediated deoxyuridine triphosphate nick-end labeling (TUNEL)+ apoptotic cells (figure 3D,F). The treatment did not affect the number of α-smooth muscle actin (α-SMA)+ or platelet-derived growth factor receptor-α (PDGFRα)+ cancer-associated fibroblasts (CAFs) (online supplemental figures 12 and 13). We further examined the effects of anti-VCAM-1 antibody on immune-inflammatory cell infiltration into the tumours. Interestingly, anti-VCAM-1-injected group showed reduced infiltration of MPO+ tumour-associated neutrophils (TANs) and F4/80+ monocytes and tumour-associated macrophages (TAMs), as well as density of CD31+ tumour blood vessels in pancreatic tumours (figure 4A,B). Meanwhile, CD11b+ granulocytes, CD8+ cytotoxic T cells and CD4+ helper T cells, and CD45R+ B cells were not altered (online supplemental figure 14). These staining conditions are shown in online supplemental table 3.

Figure 3

Improved survival rate and delay of progression of malignancy in PKF mice treated with antibody to vascular cell adhesion molecule 1 (VCAM-1). (A) Experimental setup: PKF mice were intraperitoneally injected with 50 µg of rat monoclonal antibody to VCAM-1 or isotype control immunoglobulin G (IgG), fives times per week, starting from 4 weeks old until the endpoint. (B) Kaplan-Meier survival analysis of PKF mice with administration of anti-VCAM-1 antibody or control IgG (n=5, four male and one female in both groups). p<0.01 by log-rank test. (C) Pancreatic weight (ratio to body weight) of PKF mice treated with anti-VCAM-1 antibody or control IgG. Median value is shown. **p<0.01. (D) Histological phenotypes of the pancreatic tumours by H&E staining, immunohistochemistry for Ki-67 and TUNEL (×100). Arrows indicate TUNEL+ cells. Insets show representative images of positive staining in the tumour. Scale bars: 100 µm. (E) Histological status of pancreatic tumours. Mean value and SD are shown. (F) Quantification of data presented in (D). Both groups contain six mice in (C)–(F) (Control group: three male and three female; anti-VCAM-1 antibody group: four male and two female). Median value is shown. *p<0.05. TUNEL, terminal deoxynucleotidyl transferase-mediated deoxyuridine triphosphate nick-end labeling.

Figure 4

Decrease of inflammatory cell infiltration and angiogenesis in pancreatic tumour by injection of anti-vascular cell adhesion molecule 1 (anti-VCAM-1) antibody. (A) Immunohistochemistry of PKF tumours treated with control immunoglobulin G (IgG) or anti-VCAM-1 antibody. Myeloperoxidase (MPO)+ tumour-associated neutrophils (TANs), F4/80+ tumour-associated monocytes/macrophages and CD31+ blood vessels in pancreatic tumours (×100). Insets show representative images of positive staining in the tumour. Scale bars: 100 µm. (B) Quantification of the data shown in (A). Both groups contain six mice (Control group: three male and three female; anti-VCAM-1 antibody group: four male and two female). Median value is shown. *p<0.05; **p<0.01.

We tried to determine whether blocking VCAM-1 has inhibited the onset of CAT in PKF mice. When treated with VCAM-1-neutralising antibody starting at 7 weeks of age (online supplemental figure 15A), CAT was observed in four out of six mice (67%), while it was detected in six out of six mice (100%) in the control IgG-injected group (online supplemental figure 15B). In the control group, thrombus was seen in cardiac ventricular wall, IVC, portal, splenic and pancreatic veins. In VCAM-1 antibody-treated mice, thrombus was detected in intrahepatic portal vein, pancreatic vein and IVC but not in the heart. In this cohort, anti-VCAM-1 antibody injection did not significantly reduce the weight of heart (online supplemental figure 15C). These results indicated that blockade of VCAM-1 tended to reduce the frequency of CAT and inhibited the systemic distribution of CAT, especially at the ventricular walls.

VCAM-1/sVCAM-1 production from PDAC cells is regulated by TGF-β signalling but not by anticoagulant heparin

Considering frequent CAT in PKF mice, but not in KPC or KPCflox mice, CAT might be associated with the impaired TGF-β signalling, possibly through increasing VCAM-1 production. Consistent with this idea, we have previously reported that Tgfbr2-deficient PKF cells were producing significantly higher level of sVCAM-1 compared with Tgfbr2-intact pancreatic tumour cells derived from Ptf1acre/+;LSL-KrasG12D/+ mice.19 In addition, we confirmed significantly higher Vcam-1 mRNA expression in Tgfbr2-deficient PKF cells compared with Tgfbr2-intact KPCflox cells and downregulation of Vcam-1 in Tgfbr2-intact KPCflox cells treated by TGF-β (online supplemental figure 16A,B). These results indicated VCAM-1/sVCAM-1 regulation by TGF-β signalling.

Anticoagulants, which are clinically in use to treat CAT, have been reported to reduce the incidence of cancer32 and also to inhibit tumour growth and metastasis in preclinical pancreatic cancer models.33 Thus, we next examined whether Vcam-1 expression in PDAC cells was regulated by anticoagulant heparin, but Vcam-1 expression did not change in PKF and KPCflox cells (online supplemental figure 16C), indicating that the anti-CAT and antitumour effects of anticoagulants were independent of VCAM-1 expression.

Plasma sVCAM-1 and ANP levels in human PDAC patients with CAT are increased

To determine whether the findings in PKF mice are applicable to human PDAC patients with CAT, we sequentially measured plasma levels of sVCAM-1 and ANP in PDAC patients who received chemotherapies. The plasma levels of the patients with CAT were analysed at three points: when clinically diagnosed as PDAC (pre-chemotherapy) and after the chemotherapy started, within 1 month before and after onset of CAT (pre-CAT and post-CAT, respectively). Interestingly, plasma levels of sVCAM-1 and ANP in PDAC patients with CAT were higher at pre-chemotherapy and maintained at the pre-CAT and post-CAT time points than those of PDAC patients without CAT, who have received chemotherapies for a while (19–52 months) (figure 5A,B). The sVCAM-1 level showed a tendency of slight increase at pre-onset of CAT compared with pre-chemotherapy, whereas slight decrease after the onset of CAT (figure 5A). Meanwhile, plasma ANP level tended to slightly increase in PDAC patients after the onset of CAT (figure 5B), which was similar to the results of PKF mice with and without CAT (figures 1D and 2A). Indeed, positive correlation between plasma sVCAM-1 and ANP levels in PDAC patients with CAT was observed (figure 5C). Finally, we investigated CAT in the autopsy samples of patients with PDAC. Consistent with the results in PKF mice, PDAC tumours expressed VCAM-1 (online supplemental figure 17A), indicating the PDAC cells as a source of sVCAM-1. In addition, sVCAM-1 was detected in aggregation of blood, fresh thrombi and organising thrombi, while the aggregation and thrombi were positive for vWF, PTAH and focal MT staining (online supplemental figures 17B and 18), again indicating that the thrombi might be initially aggregated platelets and then alter to fibrinous thrombi. Leucocytes were detected in the adherent lesion of fresh thrombi on the thickened endothelial intima (figure 5D and online supplemental figure 18). The leucocytes included immature CD11b+ granulocytes, MPO+ neutrophils and CD68+ macrophages, whereas pan-cytokeratin+ PDAC cells were not detected in the lesions (figure 5D), which was similar to the results from PKF mice. These staining conditions are shown in online supplemental table 4. To determine the frequency of cardiomegaly, we reviewed available autopsy records of patients with PDAC included in this study. Heart weight was significantly heavier in patients with CAT (n=6) than those without CAT (n=9); however, the ratio of heart to body weight did not show significant difference (online supplemental figure 19). Taken together, plasma sVCAM-1 and ANP levels might be useful clinical risk markers of human CAT, while VCAM-1/sVCAM-1 could be promoting aggregation of blood and onset of thrombosis in patients with PDAC.

Figure 5

Elevated plasma levels of soluble vascular cell adhesion molecule 1 (sVCAM-1) and atrial natriuretic peptide (ANP) in pancreatic cancer patients with CAT. (A) Plasma level of sVCAM-1 in pancreatic ductal adenocarcinoma (PDAC) patients without CAT (non-CAT, 6 male and 2 female), with CAT at pre-chemotherapy (Pre-chemo, 10 male and 7 female), at pre-onset of CAT (Pre-CAT, 8 male and 7 female) or post-onset of CAT (Post-CAT, 2 male and 2 female). Mean value and SD are shown. **p<0.01. (B) Plasma level of ANP in PDAC patients without CAT (non-CAT, six male and two female), with CAT at pre-chemotherapy (Pre-chemo, six male and six female), at pre-onset of CAT (Pre-CAT, six male and five female) or post-onset of CAT (Post-CAT, two male and two female). Mean value and SD are shown. **p<0.01. (C) Correlation of plasma sVCAM-1 and ANP levels in PDAC patients with CAT (PDAC-CAT, red) or without CAT (PDAC-non-CAT, blue). Approximation formula and line of each group are shown. p<0.05 by Mann-Whitney U test. (D) H&E (left: ×100 and mid: ×200) and immunohistochemical (×200) analysis of CAT in patients with PDAC using autopsy samples. Arrows indicate leucocytes in H&E staining and stained cells in immunohistochemistry. Dot lines: margin of fresh thrombus (right side) and thickened endothelial intima (left side). Insets show representative images of positive staining. Scale bars: 100 µm.


Several mouse models of CAT have been reported,34 but most of them were xenograft or allograft models. Genetically engineered murine PDAC models are superior to the grafted models especially in terms of intact tumour microenvironment, although in the most widely used KPC17 and KPCflox 29 models, CAT has been uncommon (online supplemental table 2). In contrast, CAT was frequently and systemically observed in the PKF mice not only in the veins but also in the arteries and cardiac ventricles. No microthrombus was detected in glomerular and pulmonary capillaries in PKF mice, indicating that pathogenesis of the CAT could be different from disseminated intravascular coagulation. Deep venous thrombosis (DVT) in veins around gastrocnemius muscles and cerebral infarction like Trousseau’s syndrome were not detected in PKF mice. Meanwhile, arterial CAT in the aorta and cardiac ventricles is rare in cancer patients with CAT. It might have resulted from the functional difference such as bipedalism and quadrupedalism. We observed organising thrombi in most cases of PKF mice, not only fresh thrombi, and thrombus was hardly observed in other PDAC models, indicating that the thrombosis was not just near-death or after-death events, but a characteristic of PKF mice. Although we should care about differences from the clinical conditions, PKF mice would be a unique and useful spontaneous model of PDAC-associated thrombosis/thromboembolism.

In the present study, we first report the correlation between CAT and high level of ANP. ANP, secreted from the cardiac atria, is a well-known biomarker of advanced cardiomyopathic conditions such as congestive heart failure.35 36 Arterial thromboembolism could induce hypervolaemic cardiomegaly, and pulmonary embolism resulted from DVT or thrombosis on IVC induces cor pulmonale with dilation of right ventricle and hypertrophy of right ventricular wall. Therefore, arterial and/or venous thromboembolism could increase the atrial wall stretching, thereby leading to increase of ANP. Strikingly, we found that increase of plasma ANP in PDAC patients with CAT was already detected in 9 out of 11 (81.8%) cases before clinical diagnosis of CAT and maintained during chemotherapy. In contrast, increase of plasma ANP was not detected in any cases of PDAC patients without CAT (0 out of 8; 0%) during chemotherapy. Therefore, plasma level of ANP would be a risk marker of CAT in patients with PDAC, except in patients with specific heart diseases.

Inflammatory cytokines are closely related to the pathogenesis of DVT.37 Platelet-derived IL-1β induces endothelial secretion of IL-6 and IL-8,38 interactions between platelets and endothelial cells further induce expression of ICAM-1, VCAM-1, E-selectin and P-selectin and release of those cytokines as well as TF. Indeed, increase of those soluble adhesion molecules is observed in patients with DVT.30 Similar to patients with DVT, high levels of the inflammatory cytokines including ILs and adhesion molecules were detected in PKF mice, although the sample size in the cytokine array was small (online supplemental figures 2–11). Notably, many of these cytokines and molecules including sVCAM-1 tended to be slightly lower in the PKF with CAT compared with those without CAT. It seems consistent with the plasma sVCAM-1 data of patients with PDAC: a slight increase at the pre-CAT and a slight decrease at the post-CAT (figure 5A,B). Considering that VCAM-1 was abundantly observed in the thrombi of PKF and patients with PDAC, sVCAM-1 and other molecules might have been consumed for the thrombus formation, although it is possible that other cytokines than sVCAM-1 might also be involved in thrombus formation.

It is reported that plasma sVCAM-1 level in chronic lymphocytic leukaemia patients with DVT is significantly higher than that without DVT after chemotherapy,39 which is in line with our results, indicating that the mechanism is not specific for PDAC. On the other hand, in patients with metastatic breast cancer receiving chemotherapies, plasma sVCAM-1 level was not correlated with CAT,40 suggesting that the underlying mechanisms of CAT are various depending on the cancer types.

It is known that PDAC microenvironment is highly inflammatory involving multiple types of immune cells and the inflammation accelerates Kras-mutant PDAC formation and progression.41–43 Our results suggested that VCAM-1 may be a key molecule to attract monocytes/macrophages and neutrophils into the tumour microenvironment. It has recently been reported that tumour-associated macrophages promote expression of VCAM-1 on pancreatic tumour cells,27 which suggests that there might be a feed-forward loop of macrophage/neutrophil infiltration and VCAM-1/sVCAM-1 production in the tumour microenvironment to promote PDAC development. This study indicated that blocking VCAM-1/sVCAM-1 might have reduced infiltration of immune-inflammatory cells, including TANs and TAMs, into the pancreatic tumour, reduced tumour angiogenesis, thereby leading to the delayed progression of PDAC (figure 6A). However, we could not completely exclude the possibility of other mechanism for tumour progression in this setting. More detailed analyses of immune-inflammatory tumour microenvironment as well as linkage of inflammation and tumour progression remain to be investigated.

Figure 6

A schematic diagram of hypothetical mechanism of cancer-associated thrombosis/thromboembolism (CAT) in pancreatic ductal adenocarcinoma (PDAC). (A) Effect of anti-vascular cell adhesion molecule 1 (anti-VCAM-1) antibody on pancreatic tumour progression. In pancreatic tumour microenvironment, administration of anti-VCAM-1 antibody reduces local infiltration of inflammatory cells including myeloperoxidase (MPO)+ tumour-associated neutrophils (TANs) and F4/80+ monocytes/tumour-associated macrophages (TAMs). Inhibition of the inflammation might delay tumour angiogenesis and PanINs and PDAC progression. (B) Effect of anti-VCAM-1 antibody on PDAC-associated thrombosis progression. PDAC cells and CAFs produce VCAM-1 and release sVCAM-1. Endothelium, activated by tumour-associated cytokines including interleukin-1 (IL-1) and tumour necrosis factor (TNF), expresses adhesion molecules including VCAM-1 on the surface.40 CAF secretes P-selectin (P-Sel) and activates platelet (Pt), then Pt activates leucocytes. The activated platelets and leucocytes attach on the activated endothelial cells. Meanwhile, sVCAM-1, derived from PDAC and CAF, might bind to the ligands VLA-4 on leucocytes and promote aggregation of blood and the subsequent thrombosis. Anti-VCAM-1 antibody might block sVCAM-1 and VCAM-1 and inhibit thrombosis.

Leucocytes such as neutrophils, monocytes/macrophages and MDSCs were observed in the adherent lesions of thrombi on the endothelium in PKF mice with CAT, suggesting their essential roles in the thrombus formation. Since it is known that VCAM-1 mediates leucocytes adhesion to the endothelium,21 such adhesion might mediate initial recruitment of leucocytes to the thrombus. Neutralising VCAM-1 reduced the distribution of thrombi in PKF mice, which might be due to the inhibitory effect on the initial recruitment of leucocytes to the thrombus (figure 6B). In addition, as blocking VCAM-1 reduced tumour volume in PKF mice, the effect on the thrombosis might be partially due to the reduction of tumour volume, where the VCAM-1/sVCAM-1 production was reduced, thereby inhibiting CAT progression. Although the VCAM-1-neutralising antibody treatment could not completely prevent CAT, it might be due to the later start and lower frequency of the injection than those in the survival study, or there might be other molecules than VCAM-1 involved in CAT. For example, neutrophil extracellular traps (NETs), extracellular fiber-like DNAs, were present in arterial thrombi, while administration of DNase I inhibited venous thrombosis in the xenografted mice.44

We originally reported that the PKF mice quickly die due to the tumour burden, demonstrating cachexia.18 This study also revealed frequent CAT in the PKF mice. In the observation, sometimes widespread thrombosis caused sudden paraplegia, which allowed us to predict death of mice within 24 hours, indicating the contribution of CAT to the poor prognosis. Both tumour burden and CAT can be the cause of death; however, considering that not all mice develop thrombosis at the endpoint, it is suggested that tumour burden might be the main cause of death and the thrombosis might add the last step over. The reduced tumour volume and inhibited CAT progression by treating with anti-VCAM-1 antibody might also support the idea of both contributing to the prognosis.

Alteration in TGF-β signalling might be associated with CAT, possibly through more VCAM-1/sVCAM-1 production from PDAC cells. Since it has been reported that PDAC patients with loss of SMAD4, a central mediator of TGF-β signalling, have more metastasis and poorer prognosis,45 46 CAT might be one of the underlying mechanisms of promoting metastasis and worsening the prognosis in those patients. Patmore et al described that vWF-associated coagulopathies might be required for tumour metastasis,47 suggesting that CAT might induce metastasis. Unfortunately, PKF mice die too quickly to form distant metastasis18; therefore, it was difficult to analyse the association of CAT and metastasis in this study. Meanwhile, it is recently reported that PDAC cells with intact TGF-β signalling would become more aggressive through enrichment of epithelial-mesenchymal transition-related gene expression.48 49 Taken together, association of TGF-β signalling, CAT and metastasis as well as prognosis of PDAC remains to be investigated in detail in the future.

Clinically, anticoagulant drugs are commonly used to treat CAT; however, there is a risk of life-threatening haemorrhage and currently little impact on the prognosis of PDAC.5 7 This study indicated that anticoagulant might not directly regulate VCAM-1 expression, suggesting different modes of action from VCAM-1 blockade. According to the survival extension we observed by more than four times, blocking VCAM-1/sVCAM-1 might have a strong therapeutic impact for PDAC.

We should note small sample size of the survival analysis (n=5 in each group) and the cytokine array (n=3 to 4 in each group) as the limitations in this study. In addition, the sex distribution in each experiment might have affected the results as a possible confounding factor.

In summary, this study demonstrates, using a spontaneous CAT model of PDAC mice, that VCAM-1/sVCAM-1 are highly involved in the pathogenesis of pancreatic cancer, promoting immune-inflammatory cell infiltration and tumour angiogenesis as well as systemic progression of CAT. Plasma sVCAM-1 and ANP might be potent biomarkers for predicting high-risk group and early detection of CAT in PDAC, and blocking VCAM-1/sVCAM-1 can be a potential therapeutic approach that can contribute to the prognosis of the most intractable cancer.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. All data relevant to this study are included in the article and supplementary data.If you need to contact us, email address

Ethics statements


We thank Christopher VE Wright (Vanderbilt University) for Ptf1acre/+ mice, Tyler Jacks (Massachusetts Institute of Technology) for LSL-KrasG12D/+ mice, Takuji Sato and Takumi Kawabe (CanBas, Numazu, Japan) for technical assistance and laboratory members for helpful discussions.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Related Data


  • MS and RT contributed equally.

  • Correction notice This article has been corrected since it published Online First. The author name, Jinsuk Kim, has been corrected.

  • Contributors MS, RT, HIj and SM designed experiments. MS and RT acquired and analysed data. MS, RT and HIj wrote the manuscript. TY, KM, GK and JK performed a part of experiments. KI, SM, YN and HIs provided clinical concept. YMa, YMo, MT and TU supported pathological analysis. KM, HK, HF, TN, YT, KT and YI discussed data and provided intellectual input. HLM reviewed the manuscript and KK supervised the study.

  • Funding This work was supported by the Japanese Society for the Promotion of Science Kakenhi Grant 26430107 and 17K07155, and the Fugaku Trust for Medicinal Research.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.