J. Pathol. 187: 291–294 (1999)

URINARY TISSUE FACTOR LEVELS IN PATIENTSWITH BREAST AND COLORECTAL CANCER

. 1*, 1 . 2

1University Department of Haematology, Southampton University Hospitals, Southampton, U.K.2Hemostasis and Thrombosis Research Unit, Walt Disney Memorial Cancer Institute at Florida Hospital, Altamonte Spring,

FL 32701, U.S.A.

SUMMARY

Activation of blood coagulation is a common complication of cancer in man and experimental animals. The causes of such activationmay be multifactorial, but increased production of tissue factor (TF) by the host mononuclear cells may be involved. TF is not onlyproduced by human monocytes (mTF) and tumour cells, but is also found in urine (uTF), where measurements might be clinicallyimportant. Using a highly reproducible (intra-assay CV 2·3 per cent and inter-assay CV 8·1 per cent) one-stage kinetic chromogenicassay (KCA) developed by this group, uTF levels were measured in controls [healthy volunteers (n=57), patients with renal stones anda normal ESR (n=30)] and in patients with benign and malignant diseases of the breast (n=94) and large bowel (n=62). Each benigndisease group was sub-divided into inflammatory and non-inflammatory categories. There were no significant differences between thecontrols and the benign non-inflammatory groups, so they were unified for further analysis. Malignant groups, irrespective of tumourtypes, showed significantly higher uTF levels than controls (p<0·001 for breast and p<0·01 for large bowel). Similarly, breast andcolorectal benign inflammatory groups showed significant increases over controls (p<0·01 and p<0·001, respectively). Patients withmalignant disease showed uTF activity above the upper quartile range of the normal control group for breast, 77·3 per cent, and largebowel, 73 per cent. uTF levels were related to histological tumour grading and were higher in non-surviving patients. In conclusion, uTFlevels are raised in malignant and inflammatory disease compared with controls and patients with non-inflammatory conditions. uTFlevels may reflect tumour progression. Copyright ? 1999 John Wiley & Sons, Ltd.

KEY WORDS—urinary tissue factor; coagulopathy; cancer

INTRODUCTION

The association between cancer and altered haemos-tasis is well established and TF is implicated.1,2 Inparticular, there is good evidence to support the involve-ment of mTF and/or macrophages in inappropriateclotting activation. Measurement of mTF may be diag-nostically useful in cancer and inflammatory conditions,but is technically demanding and difficult to adapt in aclinical laboratory environment. Urine is known tohave a powerful procoagulant activity (PCA)3 thatnormalizes the clotting time of haemophiliac patientsin vivo.4 The PCA was first thought to be a tissuethromboplastin-related substance.5 Subsequently,Wiggins et al. showed that urinary PCA resided onlipid-associated vesicles and was mainly Factor VII-dependent, as assessed by clotting assays in humanFactor VII-deficient plasma.6 Recently, Carty et al.confirmed that the urinary PCA was TF, by demonstrat-ing almost total inhibition of the activity by a specificantibody to human TF.7,8

TF apoprotein is a single-chain, integral membrane-bound glycoprotein with no intrinsic proteaseactivity.9,10 It serves as a receptor and essential cofactorfor the serine protease blood coagulation Factors VIIand VIIa in the activation of Factors X and IX.11 Thus,TF is an important initiator of blood coagulation.12

Evaluation of uTF in malignancy is currently the subject

CCC 0022–3417/99/030291–04 $17.50Copyright ? 1999 John Wiley & Sons, Ltd.

of considerable interest. Increased uTF activity isreported in patients with colorectal cancer, inflamma-tory bowel disease, and breast cancer,7,8 and in patientswith transitional cell carcinoma of the bladder andprostate, compared with controls and patients withbenign prostatic hypertrophy.13,14

Although uTF measurements may be diagnosticallyuseful the assay system used in the above studies maynot have been completely reliable. Recently, we havedeveloped a highly reproducible uTF assay (intra-assayCV 2·3 per cent and inter-assay CV 8·1 per cent) whichwe have applied to patients with glomerulonephritis.15

In the present study, we have examined the applicationof this assay to patients with breast and colorectalcancer and have compared the results with thoseobtained in subjects with benign disease and in normalcontrols. The effect of inflammatory conditions on uTFPCA was studied. uTF levels were also correlated withhistological tumour grading and patient survival.

MATERIALS AND METHODS

*Correspondence to: Bashir A. Lwaleed, University Department ofHaematology, Level F (827), South Academic Block, SouthamptonGeneral Hospital, Tremona Road, Southampton SO16 6YD, U.K.

Controls and patientsA total of 243 subjects were studied. The study

groups’ anthropometric data are recorded in Table I.Ethical committee approval was obtained for the studyand informed consent was sought from each patient onadmission to the surgical wards or attending the RenalMedical Outpatient Clinic at Southampton UniversityHospitals. Urine samples were obtained prior tooperation. All patients had been clinically diagnosed and

Received 2 March 1998Revised 2 June 1998

Accepted 7 August 1998

292 B. A. LWALEED ET AL.

the histopathology reports were subsequently obtained.Tumours were classified according to the World HealthOrganization classification for breast cancer and Dukes’categorization for colorectal cancer.

Urinary tissue factor measurements

A random midstream urine sample was collected fromeach subject into sterile universal containers withoutpreservative. Samples were then sedimented, solubilized,and assayed using a one-stage KCA. Briefly, theextracted uTF, in the presence of recombinant FactorVIIa (rVIIa) and Ca2+, forms a complex (TF:rVII-a:Ca2+) which then directly activates Factor X to FactorXa. Generation of Factor Xa, which is proportional tothe amount of TF in the urine sample, is determined bymeasuring its action on a Factor Xa-specific chromo-genic substrate. The rate of the reaction at 405 n isdetermined in a Biokinetics EL-312e micro-plate reader(Bio Tek Instruments, Inc., Winooski, VT, U.S.A.),programmed to read the absorbance at 30 s intervalsover a period of 35 min. Absorbance values were con-verted to TF (ng/ml) using a calibration curve con-structed from serial dilutions of recombinant relipidatedTF (rrTF, 0·4–83 ng/ml), prepared according to Carsonand Konigsberg16 in BOG (â-octyl-glucopyranoside;Sigma Chemical Company, Poole, Dorset, U.K.). Formaximum accuracy, the calibration standards weremeasured in each plate.

Expression of the urinary tissue factor results

The results (ng/ml) were corrected for the dilution ofthe urine using the creatinine concentration of thesample. Final results were expressed as uTF ng/ml permg creatinine [uTF (ng/ml)/creatinine (mg/dl)#100].

Statistical analysis

Data were included in a database and analysed bythe STATGRAPHICS= statistical software system.Data were not normally distributed and summarystatistics were expressed as medians and interquartileranges (IQR). Differences between two groups wereassessed by the Mann–Whitney U-test. Differences intumour grading were tested by one-way analysis ofvariance (ANOVA).

Copyright ? 1999 John Wiley & Sons, Ltd.

RESULTS

Table I—Sample size of urinary tissue factor activity, median and range, age and sex for the groups studied

Group n

Medianage

(years)

Agerange(years)

Male(n)

Female(n)

Normals 57 33 5–69 30 27Renal stones 30 44 20–88 14 16Benign breast 31 47 20–76 — 31Breast cancer 63 61 26–87 — 63Benign large bowel 39 64 39–92 22 17Large bowel cancer 23 73 43–96 10 13

Fig. 1—Distribution of urinary tissue factor (ng/ml) in normal con-trols, renal stone, non-inflammatory, and inflammatory benign diseaseand malignant disease of the breast

Fig. 2—Distribution of urinary tissue factor (ng/ml) in normalcontrols, renal stones, non-inflammatory, and inflammatory benigndisease and malignant colorectal disease

Assessment of urinary tissue factor activity in stratifiedbenign neoplasia and malignant disease

The median and IQR of the uTF activity for thesegroups are shown in Figs 1 and 2. The normal controlgroup displayed a very narrow range compared with theother groups.

In breast disease (Fig. 1), no significant difference wasobserved between the normal controls, the renal stones,

J. Pathol. 187: 291–294 (1999)

293URINARY TISSUE FACTOR LEVELS IN CANCER

and the non-inflammatory benign breast lesions, but themalignant group showed significantly higher uTF levelsthan the control groups (p<0·001). uTF levels were alsosignificantly raised in the inflammatory benign diseasecompared with all controls (p<0·01). No significantdifferences were observed between the malignant and theinflammatory groups.

In the colorectal group, again no significant differencewas seen among control groups (Fig. 2). The inflamma-tory benign disease group showed a significant increaseover controls (p<0·001). The malignant group showed asignificant increase compared with the control groups(p<0·01), but no significant differences were observedbetween the malignant and the benign inflammatorygroups.

uTF activity above the upper quartile ranges of thenormal control group was found in 77·3 and 73·0 percent of patients with breast and colorectal malignancy,respectively.

Tumour grade

Tumour grading was based on the histological tumourgrading. There was an increase in uTF corresponding toa higher tumour grade (Table II), but this increase failedto reach statistical significance (ANOVA; p>0·05 in eachcase).

Patient survival

There was a tendency for an increase in the uTF levelsin non-surviving patients compared with survivors, irre-spective of tumour type (data not shown), but thedifference between those patients who survived andthose who died was not statistically significant (p>0·05).

Table II—Urinary tissue factor activity (ng/ml) and histologi-cal tumour grades in patients with breast and colorectal cancer

Grade n Median IQR

Breast 631 13 12·0 8·0–18·02 20 18·0 11·0–23·03 30 20·0 15·0–27·0

Colorectal 23Dukes’ A 5 12·0 7·9–15·0Dukes’ B 7 14·0 7·0–24·0Dukes’ C 11 17·0 10·0–25·0

DISCUSSION

In western society, cancer remains second to coronaryheart disease as the commonest cause of death and thesearch for markers of the presence and progression ofmalignant disease continues. The evaluation of urinarymaterials as cancer markers, has long been seen asattractive. Mundy studied the value of the urinaryhydroxyproline level in prostatic carcinoma.17 Therelationship between urinary fibrinogen degradation

Copyright ? 1999 John Wiley & Sons, Ltd.

products and active bladder cancer was investigated byWajsman et al.18 and Martinez-Pinerio et al.,19 and agood correlation was observed. Carty et al.7,8 andAdamson et al.13,14 demonstrated increased levels ofuTF in patients with cancer and inflammatory boweldisease, but the assay system used in the latter studieswas not completely reliable. The development and vali-dation of a simple and clinically applicable KCA formeasuring uTF activity would therefore be desirable.

We have developed a highly reproducible assay foruTF measurement which is not significantly affected byage, gender or smoking habits (to be published). Amajor advance of the new assay is the smaller variationwithin the study groups, particularly with normal con-trols, which exhibited a bell-shaped distribution allow-ing better separation of the groups. This assay has beenpreviously used to assess glomerular damage15 and datasuggested that uTF concentrations in this group mayreflect the pathogenesis of glomerular damage, or itsdegree. The application of this assay to patients withbreast and colorectal disease is reported in the presentstudy.

There was no significant difference between the con-trol groups and those with benign non-inflammatoryconditions of the breast and colorectum, whereas asignificant difference was observed between the controlsand those with benign inflammatory conditions, whichincluded mastitis, mammary duct ectasia, inflammatorybowel disease, and diverticulitis. Spillert and Lazaro andOsterud et al. suggested that inflammatory conditionsinterfere with the host immune response, which leads toan increase in mTF expression,20,21 a mechanism whichcould similarly explain the elevation of uTF in theseconditions. Each malignant disease group showed sig-nificant differences from the control groups and theappropriate organ-specific non-inflammatory diseases,but not from the relevant inflammatory benign disease.The same results were obtained by Dasmahapatra et al.,using whole blood re-calcification time22 and Contrinoet al., who found increased expression of TF in endo-thelial cells of the blood vessels of malignant breasttumours but not of the benign lesions, using bothimmunohistochemical techniques and a novel probe forfunctional TF.23 These observations suggest a diagnosticrole for uTF assay in malignant diseases, in the absenceof extensive inflammation.

The malignant and the benign inflammatory groupsshowed wider ranges in uTF levels than that of controls.This could be attributed to variation in tumour gradeand type. The malignant group included patients withorgan-confined cancer and those with extensive localdisease and/or lymph node metastasis. Previous studiesof uTF activity have failed to differentiate betweentumour grade and uTF level and no difference wasfound between patients with superficial tumours andthose with invasive cancer.13,14 In addition to theheterogeneity of the study groups, the assay’s technicalaspects may also have contributed to the lack of dis-crimination in these results. In the present study, therewas a trend, which did not reach statistical significance,towards increasing uTF levels corresponding to tumourprogression as determined by histological tumour

J. Pathol. 187: 291–294 (1999)

294 B. A. LWALEED ET AL.

grading and levels were higher in patients who sub-sequently died.

Patients with benign disease were also heterogeneousand comprised various inflammatory conditions that areassociated with increased mTF expression.24 This couldaccount for the wider ranges obtained in this group.Indeed, when the benign groups were sub-divided intoorgan-specific non-inflammatory and inflammatorybenign diseases, the within-group variation was signifi-cantly reduced in the non-inflammatory compared withthe inflammatory diseases. This was particularly seen forthe colorectal group, where inflammation is common.The renal stone group, which was included as a non-inflammatory control, also showed a very low variationcompared with the inflammatory groups. However,Carty et al. found that subjects with rheumatoid arthri-tis, all of whom had active disease, demonstrated normaluTF levels,7,8 suggesting that uTF does not behavesimply as one of the acute phase reactants.

uTF levels are significantly increased in the malignanttumours studied as well as the benign inflammatorydisease, when compared with control or benign non-inflammatory conditions, although there was a slightoverlap between these groups and the control groups.This overlap can partly be attributed to the heterogen-eity of the benign controls and the malignant groups, asdiscussed earlier; medications at the time of samplecollection; and the presence of unknown blood-derivedand/or urinary inhibitors to uTF activity.

When cancer patients were classified according to theorgan affected, a total increase in uTF activity wasobserved for colorectal cancer. Similar results have beenreported by others.7 The mechanisms of uTF elevationremain to be resolved and it is, as yet, not known whateffect tumour burden has on uTF levels, but it is possiblethat colorectal cancer tissue may stimulate peripheralblood monocytes or tissue macrophages to secrete morePCA. Inflammatory conditions frequently associatedwith colorectal cancer may also play a part, sinceisolated monocytes from patients with Crohn’s diseasehave been shown to express increased levels of TF whichcorrelated with disease activity.24 Underlying bacterialinfection is a common occurrence in colorectal cancerand a few patients with urinary tract infection showedincreased uTF levels.25 Colorectal cancer expresseselevated levels of cellular TF,26 and higher levels of freeTF were observed for colorectal tissues than for breastcancer.27 In this context, the majority of colon cancerpatients present at later stages of disease progression,particularly relative to breast cancer, which is subject toscreening. Furthermore, a particular tumour may acti-vate coagulation by either a single or multiple mecha-nisms28 and colorectal cancer may activate coagulationin different ways.

In conclusion, the data presented here suggest that theuse of the new uTF assay has markedly improveddiscrimination between the study groups. The test candistinguish patients with malignancy from normal con-trols or benign non-inflammatory diseases in the absenceof inflammation. In addition, uTF levels were higher inpatients with a poorer prognosis and those who subse-

Copyright ? 1999 John Wiley & Sons, Ltd.

quently died. The relationship between uTF and histo-logically defined disease stage suggests a potential rolefor uTF in assessing tumour progression and response totreatment. Thus, uTF levels may play a role in screeningpatients who are at risk of and most likely to benefitfrom detailed investigation.

REFERENCES1. Rodgers GM. Haemostatic properties of normal and perturbed vascular

cells. FASEB J 1988; 2: 116–123.2. Cybulsky MI, Chan MKW, Movat HZ. Acute inflammation and micro-

thrombosis induced by endotoxin, interleukin-1, and tumour necrosis factorand their implications in Gram-negative infection. Lab Invest 1988; 58:365–378.

3. Grunke W. Studien über die Bluterinnung mit besonderer Berücksichtigungder Haemophilie. I. Mitteilung: Einfluss des Harnes auf die Gerinnunghaemophilen Blutes. Blutes Z Ges Exp Med 1835; 96: 512–516.

4. Tocantins LM, Lindquist JN. Thromboplastic activity of the urine. ProcSoc Exp Biol NY 1947; 65: 44–49.

5. von Kaulla KN. Extraction and concentration of thromboplastic materialfrom human urine. Proc Soc Exp Biol Med NY 1956; 91: 543–545.

6. Wiggins R, Glatfelter A, Kshirsagar B, Beals T. Lipid vesicles and theirassociation with procoagulant activity and glomeruli of rabbits withnephrotoxic nephritis. Lab Invest 1987; 56: 264–272.

7. Carty N, Taylor I, Roath OS, El-Baruni K, Francis JL. Urinary tissuefactor activity in malignancy. Thromb Res 1990; 57: 473–478.

8. Carty N, Taylor I, Roath OS, El-Baruni K, Francis JL. Urinary tissuefactor activity in colorectal disease. Br J Surg 1990; 77: 1091–1094.

9. Bach RR, Nemrson Y, Konigsberg WK. Purification and characterizationof bovine tissue factor. J Biol Chem 1981; 256: 8324–8331.

10. Guha A, Bach RR, Konigsberg W, Nemerson Y. Affinity purification ofhuman tissue factor. Interaction of Factor VII and tissue factor in deter-gents micelles. Proc Natl Acad Sci USA 1986; 83: 299–302.

11. Nemerson Y, Bach R. Tissue factor revisited. Prog Hemost Thromb 1982; 6:237–261.

12. Nemerson Y. The tissue factor pathway of blood coagulation. SeminHaemost 1992; 29: 170–176.

13. Adamson AS, Francis JL, Roath OS, Witherow RO’N, Snell ME. Urinarytissue factor levels in transitional cell carcinoma of the bladder. J Urol 1992;148: 449–452.

14. Adamson AS, Francis JL, Witherow RO’N, Snell ME. Urinary tissue factorlevels in prostatic carcinoma: a potential marker of metastatic spread. Br JUrol 1993; 71: 587–592.

15. Lwaleed BA, Bass PS, Chisholm M, Francis JL. Urinary tissue factor levelsin glomerulonephritis: a potential marker of glomerular injury. J Clin Pathol1997; 50: 336–340.

16. Carson SD, Konigsberg WH. Cadmium increased tissue factor (coagulationfactor III) activity by facilitating its reassociation with lipids. Science 1980;208: 307–309.

17. Mundy AR. Urinary hydroxyproline excretion in carcinoma of the prostate:a comparison of 4 different modes of assessment and its role as a marker. BrJ Urol 1979; 51: 570–574.

18. Wajsman Z, Merrin CE, Chu MT, Moore RH, Murphy GB. Evaluation ofbiological markers in bladder cancer. J Urol 1975; 114: 879–883.

19. Martinez-Pinerio JA, Pertusac C, Maganto E, et al. Urinary fibrinogendegradation products (FDP) in bladder cancer. Eur Urol 1978; 4: 348–350.

20. Spillert CR, Lazaro EJ. Shortened endotoxin-activated clotting time inpatients with carcinoma. Am Surg 1987; 53: 164–166.

21. Osterud B, Olsen JO, Wilsgard L. The role of arachidonic acid release andlipoxygenase pathway in lipopolysaccharide-induced thromboplastinactivity in monocytes. Blood Coagul Fibrinol 1990; 1: 41–46.

22. Dasmahapatra KS, Cheung NK, Spillert C, Lazaro E. An assessment ofmonocytes’ procoagulant activity in patients with solid tumours. J Surg Res1987; 43: 158–163.

23. Contrino J, Hair G, Kreutzer D, Rickles FR. In situ detection of expressionof tissue factor in vascular endothelial cells: correlation with the malignantphenotype of human breast tissue. Nature Med 1996; 2: 209–215.

24. Edwards RL, Levine JB, Green R, et al. Activation of blood coagulation inCrohn’s disease: increased fibrinopeptide A levels and enhanced generationof monocytes’ tissue factor activity. Gastroenterology 1987; 92: 329–337.

25. Lwaleed BA, Bass PS, Rogerson ME, Francis JL, Chisholm M. Effect ofkidney function and disease status on urinary tissue factor measurements.J Clin Pathol 1998; 51: 234–237.

26. Szczepanski M, Bardadin K, Zawadzki J, Pypno W. Procoagulant activityof gastric, colorectal and renal-cancer is Factor-VII-dependent. J CancerRes Clin Oncol 1988; 114: 519–522.

27. El-Baruni K. Factor X-activating procoagulant in normal and malignantbreast tumour. Ph.D. thesis. Southampton University, 1990.

28. Murray JC. Coagulation and cancer. Br J Cancer 1991; 64: 422–424.

J. Pathol. 187: 291–294 (1999)