Mark T. Wimmer, MD

Residents’ Grand Rounds

March 25, 1997

Case Presentation

CC: Right lower extremity edema

HPI: AT is an 84 yo BF who developed the sudden onset of right lower extremity swelling 5 d PTA after taking a long bus trip to Ohio. Upon returning to NC, lower extremity Duplex scanning revealed bilateral deep venous thrombi in proximal lower extremities. 2 weeks PTA the patient had complained of shortness of breath and a chest X-ray then had revealed a left pleural effusion but work up of this had been deferred until after the patient’s trip.

Medications: HCTZ SH: no alcohol or tobacco

PMH: asthma, HTN Allergies: none

PE: BP 142/60 P 70 RR 16 AF

Elderly BF in NAD

Lungs- decreased BS at left base

Ext- marked swelling of RLE below knee, minimal tenderness

Breast- no masses

Rectal- heme -; no masses

Labs: WBC 3.2 Hgb 11.9 MCV 90

Ferritin 367 SMAC and LFTs WNL

CXR- mod. left sided pleural effusion

Thoracentesis- 1556 RBCs, 750 WBCs--100% monos

exudative by TP and LDH

Cytology- metastatic adenocarcinoma with signet ring cells,

possible sites of origin limited to GI tract, breast, lung, and

pancreas

Hospital Course: Pt. was treated with seven days of intravenous heparin. She was started on Coumadin the second day of her hospitalization. Further work up for malignancy including CT scan of chest and abdomen as well as colonoscopy were negative for malignancy. The patient declined further work up or treatment and was therefore discharged when her INR was >2.0 on Coumadin.

History: The relationship between deep venous thrombosis and malignancy was first described by Armand Trousseau²⁹ who, in 1865, reported a high incidence of thrombosis in patients with gastric carcinoma.²³ In 1938, Sproule²⁸, confirmed Trousseau’s earlier observations by demonstrating an increased incidence of both arterial and venous thrombosis in an autopsy series of patients who had died of various malignancies.¹² Today it is widely accepted that malignancy predisposes patients to venous thrombosis.

Incidence: It has been estimated that approximately fifteen percent of patients with malignancy will develop clinically significant thrombosis²⁷. While the incidence of venous thrombosis seems to vary somewhat depending on the type of malignancy, most types of cancer have been clearly demonstrated to place patients so afflicted at higher risk of developing thrombotic complications than age-matched controls. Sarasin et al.²⁶ report that the incidence of acute proximal deep venous thrombosis per month in hospitalized sixty to seventy year old patients with pancreatic carcinoma and gastrointestinal cancer may approach 1.5% per month, whereas age-matched hospitalized patients with nonmalignant conditions have a monthly incidence of approximately .01%. Thus, the relative risk of patients with pancreatic and gastric carcinoma of developing an acute proximal DVT is 150 times that of age matched controls. The relative risk of the development of an acute deep venous thrombosis in patients with breast and lung cancer reported in the same study²⁶ when compared to age-matched controls was somewhat lower but still significant at 4 and 90 times (respectively) that of the controls. While these estimates are based on the accumulation of data from multiple longitudinal studies and therefore must be subject to critical evaluation, there is an enormous body of literature that confirms the role of malignancy in thrombotic complications.

Etiology and Pathogenesis: Many different mechanisms have been proposed to explain the hypercoagulable state found in malignancy. Much of the information we have regarding the pathogenesis of thrombosis in cancer is based on in vitro studies and animal models. In addition to the known risk factor of prolonged immobilization, patients with malignancies have specific coagulation abnormalities which may result in chronic activation of the coagulation system.

Tumor cell procoagulants have been discovered which generally fall in one of two categories—tissue factor and cancer procoagulant.¹² Tissue factor is normally expressed on human parenchymal and connective tissue cells. Tissue factor is not under normal circumstances exposed to circulating plasma except in instances of vascular injury. Recently, the development of monoclonal antibodies directed at tissue factor along with immunohistochemical staining techniques⁶ have allowed scientists to demonstrate the presence of tissue factor in many different types of cancer cells, particularly those of epithelial cell origin.²³ Rao²³ found that all types of lung cancer, squamous cell carcinoma of the head and neck, adenocarcinomas of the colon, pancreas, and stomach, and tumors arising from the genitourinary epithelia all express tissue factor. As you recall, normal human tissue factor, when exposed to circulating plasma, complexes with factor VIIa, initiating the extrinsic pathway of the coagulation cascade. It has subsequently been proposed that exposure of tumor-generated tissue factor to circulating plasma may activate the coagulation cascade and predispose some patients with malignancy to the development of deep venous thrombosis.²³

Cancer procoagulant has also been implicated as another potential initiator of the coagulation cascade in patients with known malignancy. Cancer Procoagulant (CP) is a cysteine protease that directly activates factor X and has been recovered from the tumors of several different malignancies including those of the colon, breast, lung, kidney, and from melanoma.^10,12 CP may also be found in extracts of leukemic blast cells, particularly those of the M1-M4 subtypes of acute myelogenous leukemia.¹² Factor Xa, which is formed by CP activity, then complexes with factor Va to form the prothrombinase complex. This complex is in turn responsible for converting prothrombin to thrombin. Thrombin then cleaves fibrinogen to form fibrin—the essence of a thrombus. It is therefore believed that Cancer Procoagulant, because of its role in the activation of factor X, serves to activate the coagulation system in a way that predisposes to thrombus formation.

Other mechanisms of prothrombotic activity have also been implicated in malignancy including procoagulant activities of normal host tissues in response to tumor development. Human monocytes can produce coagulation factors, including tissue factor in response to stimulation by tumor-specific antigens, cytokines, and even endotoxin-induced inflammation.¹² It is felt that monocytes and macrophages stimulated directly or indirectly may express coagulation factors which could initiate the coagulation cascade resulting ultimately in thrombosis.

In addition to elaborating coagulation factors, monocytes as well as other cells of the immune system may liberate such cytokines as TNF and IL-1 which result in the up-regulation of leukocyte adhesion molecules on endothelial cells and the production of platelet activating factor—all of which may predispose to a prothrombotic state.¹²

Many other complex mechanisms have been recognized as contributing to the prothrombotic state of malignancy but are beyond the scope of this paper and have been reviewed extensively elsewhere.^10,12,23 Needless to say, there is probably no single explanation to the predisposition to thrombosis in malignancy, rather the hypercoagulable state is the product of many intertwined factors which are as diverse and as complex as the coagulation cascade itself.

Clinical Presentation: The clinical presentation of deep venous thrombosis in malignancy of course depends on the location of the thrombus. "Trousseau’s syndrome" classically describes a pattern of migratory, recurrent superficial thrombosis often involving the upper extremities and associated with gastric carcinomas.¹² This syndrome is quite uncommon.

The most common location of thrombosis is of the lower extremities. The signs and symptoms of lower extremity deep venous thrombosis are very nonspecific and insensitive.¹² The classic symptoms of painful swelling of the lower extremity need to be evaluated with further diagnostic tests as diagnosis by physical exam is extremely unreliable.

The presentation of patients with acute pulmonary embolism is somewhat varied. The most common complaint is the sudden onset of unexplained dyspnea. Hemoptysis and pleuritic chest pain may also be present if pulmonary infarction occurs. Other presenting symptoms may include syncope and unexplained tachyarrhythmias. The only consistent finding on physical exam is tachycardia although in rare cases of massive pulmonary embolism a palpable right ventricular lift coupled with a loud pulmonic valve closure and/or prominent jugular venous a waves may be found. Occasionally, systolic or continuous murmurs accentuated by inspiration may be heard and are associated with turbulent flow in partially obstructed pulmonary vessels.¹³ None of these signs or symptoms are specific and therefore further work up should be based on clinical suspicion of pulmonary embolism.

Upper extremity venous thrombosis has become more common since the institution of long-term indwelling venous catheters. The presentation again depends on the exact location of the thrombosis. Subclavian vein thrombosis may present clinically with swelling in the upper extremity in conjunction with neck, chest wall, or shoulder pain.¹² Dilated collateral veins may be seen on physical exam along with continued upper extremity venous distention despite elevation of the affected extremity above the level of the heart. Superior vena cava syndrome may develop from thrombosis alone and may easily be mistaken for extrinsic compression of the venous system particularly in patients that have large tumor masses involving the lung or mediastinum.¹⁹

Diagnosis: The diagnostic algorithm employed if deep venous thrombosis is suspected in patients with overlying malignancies is similar to that which would be used in any patient with suspected thrombosis, with a few caveats which will be noted below.

Several noninvasive methods have been introduced to evaluate the deep venous system of the lower extremities. Earlier methods are, however, fast losing popularity with the more recent development of Duplex scanning.

Impedance plethysmography (IPG) is a method by which blood volume changes in the calf are measured as changes in electrical resistance (impedance) with the inflation and deflation of a pneumatic thigh cuff. Blood volume changes are limited in the case of proximal venous obstruction.¹⁹ Hull et al.¹⁷, using ascending venography as a gold standard, derived a sensitivity of 93% and a specificity of 97% in detecting proximal vein thrombosis. The sensitivity for calf vein thrombosis was somewhat less at 83%. A later study¹⁵ demonstrated the safety of serial IPG studies (days 1, 2, 5, and 10) to evaluate clinically suspected deep venous thrombosis in an outpatient setting as opposed to resorting to venography in every instance. IPG, however, cannot differentiate between impaired venous return secondary to extrinsic tumor compression of the deep venous system and that associated with deep venous thrombosis and therefore false positives may result in patients with bulky pelvic or intraabdominal malignancies.¹⁹

Doppler ultrasonography which analyzes flow in the venous system and B-mode ultrasonography which produces real-time images of the deep veins were originally used as separate modalities for the diagnosis of deep venous thrombosis. Alone, these two methods are quite sensitive for detecting proximal lower extremity deep venous thrombosis.¹⁹ Duplex scanning combines both methods and is very sensitive and specific for detecting deep venous thrombosis in the population as large. A meta-analysis³¹ evaluating studies that compared duplex scanning to venography revealed a combined sensitivity of 95% for detecting proximal thrombosis and a combined specificity of 99%. It is therefore believed by some authors¹² that Duplex scanning may be used serially in the outpatient diagnosis of DVT since this modality appears to be more sensitive and specific than IPG.

Again, in patients with bulky pelvic tumors, venous flow may be obstructed by extrinsic compression, thus resulting in false positive doppler studies. Also B-mode ultrasonography is inaccurate above the inguinal ligament-- therefore one may have to resort to venography if results of noninvasive studies are equivocal. Even venography may be difficult to interpret in patients with extrinsic compression of the venous system by tumor mass. In these cases, some authors¹⁹ recommend contrasted computed tomography scans to assist in differentiating between external compression and actual venous thrombosis.

When pulmonary embolism is clinically suspected, initial work up should include PA and lateral chest radiograms, electrocardiography, and an arterial blood gas to rule out other obvious explanations for the patient’s symptoms.¹⁹ A ventilation-perfusion scan is usually the next test that should be obtained in the work up of pulmonary embolism. A high probability scan (one or greater segmental mismatched defects) generally is sufficient to rule in the diagnosis of pulmonary embolism and anticoagulation therapy may be instituted unless the patient is at high risk of hemorrhage. If the patient is at high risk, then one may want to obtain a pulmonary angiogram to confirm the diagnosis.¹⁹

The finding of a completely normal perfusion scan is usually sufficient to rule out the diagnosis of pulmonary embolism and subsequently another source of the patient’s symptoms should be sought.¹²

Unfortunately, in many cases patients will have "low probability" (segmental matched defect or subsegmental perfusion defects) or indeterminate scans (perfusion defect in the area of a radiographic abnormality). The term "low probability" is actually a misnomer because from 25-40% of patients with low probability scans will have angiographic evidence for pulmonary embolism.¹⁶ Up to 21% of patients with indeterminate scans will have a pulmonary embolism detected by angiography.¹⁶ In any patient with a malignancy, it would be somewhat risky to rule out pulmonary embolism on the basis of a non-high probability scan. In these cases it is recommended that pulmonary angiography be performed to help establish a definitive diagnosis.

In general, the physical exam is somewhat more reliable in helping to establish the diagnosis of upper extremity venous thrombosis.¹⁹ There are no controlled trials that compare noninvasive testing to venography in the case of upper extremity thrombosis. The diagnosis can be made definitively with venography, however, some noninvasive methods such as ultrasonography or computed tomography may be useful in securing a diagnosis.¹²

Therapy: The therapy for deep venous thrombosis associated with malignancy is basically the same as that for deep venous thrombosis in general. There are very few studies that concentrate specifically on anticoagulation in patients with malignancy and therefore we must extrapolate treatment strategies from the body of literature addressing the treatment of deep venous thrombosis in the population at large. Because the topic of the treatment of DVT in general is a subject unto itself and is beyond the scope of this paper, we will only briefly discuss principles of treatment below.

Initial treatment for DVT or thromboembolism should begin with heparin with the intended goal of preventing clot propagation and further thromboembolism, and to palliate symptoms. Traditionally, thrombosis has been treated with either intravenous or subcutaneous unfractionated heparin. Intravenous heparin should be administered initially as a bolus to rapidly achieve an APTT of greater than 1.5 times that of the control. A recent study has demonstrated the effectiveness of weight based heparin protocols in rapid achievement of therapeutic APTTs and this appears to translate in fewer thromboembolic complications without generating excessive risk of hemorrhage.²⁴

Very recent studies have generated excitement about the possibility of outpatient therapy for deep venous thrombosis with low-molecular-weight heparin.^18,19 The treatment with twice daily injections of subcutaneous low molecular weight heparin for at least five days with oral anticoagulation beginning on the second day appears to be at least as safe and efficacious as traditional intravenous heparin therapy in the treatment of acute, uncomplicated, proximal deep venous thrombosis. Again, no studies with low-molecular-weight heparin have been directed specifically at therapy in patients with cancer and therefore the practice must be approached with caution.

Warfarin remains the agent of choice in the long-term treatment of venous thrombosis. Heparin, regardless of the route administered, should be continued for at least five days and should not be stopped until the patient has achieved a therapeutic PT_INR of 2.0-3.0 after oral anticoagulation with warfarin.¹⁴ Warfarin may be safely started within the first 24 hours of therapy with heparin.⁸ A recent prospective cohort study³ seemingly demonstrates the relative safety of oral anticoagulation in patients with malignancy. In this study, a cohort of patients with malignancy undergoing active anticoagulation with warfarin was compared to a group of patients who were taking warfarin for various causes but did not have an underlying cancer. Both groups had similar complication rates (major bleeding, minor bleeding, or recurrent thrombosis). The incidence of major bleeding in the malignancy group (.0074 per treatment month) was statistically insignificant from that of the patients without malignancy (.0025 per treatment month). Although this study, due to its small enrollment, may not have the power to detect small differences in complication rates, it is one of the first of its kind to investigate the safety of oral anticoagulation specifically in patients with cancer.

For those patients that develop recurrent thromboembolism while therapeutic on warfarin, it is recommended by several authors^12,27 that the dose of warfarin be pushed to achieve an INR of 2.5 to 4.5 while heparin is restarted for a five day period. This recommendation lacks substantial backing in clinical trials. In such instances, placement of a vena cava filter may also be a viable option.

The length of required oral anticoagulant therapy in patients with active malignancy is controversial. Some authors advocate the traditional three month course of therapy²⁷ while others argue that it is prudent to continue therapy as long as the patient remains at high risk for venous thrombosis.¹⁹ No studies have been performed to evaluate the appropriate length of oral anticoagulation in patients with active malignancy.

Traditionally, the presence of pericardial metastases and primary brain tumors have been considered a relative contraindication to anticoagulation. However, a case series study² of patients with glial neoplasms performed in 1990 reveals a low incidence of serious complications related to long-term oral anticoagulation. More studies need to be performed to elucidate the exact complication rate in patients who are apparently at high risk for hemorrhagic complications.

Thrombolytic therapy is generally not recommended in the treatment of pulmonary embolism in patients with malignancy as it has not been studied in this patient population. Some authors believe that it should only be used in cases of massive pulmonary embolism in patients who otherwise have a good prognosis from the standpoint of the malignant process with which they are afflicted.^19,27 The exception to this rule may apply to the local infusion of thrombolytics into thrombi that have formed around permanent central venous catheters. In one study⁷, local urokinase, infused at a rate of 500 to 2000 U/kg/hr, resulted in complete lysis of 25 out of 30 thrombi; this therapy was usually more successful if performed within one week of the patient’s symptomatic presentation. Only minor bleeding was noted in five of these patients. In the study mentioned above, all of the patients received an additional week of heparin after the administration of local urokinase. In some of these patients, oral anticoagulation was instituted for three months following thrombolysis; however, those who redeveloped thrombi seemed to do so regardless of warfarin administration and the authors of the study did not routinely recommend the use of warfarin if the thrombus had resolved by the completion of a week of intravenous heparin. If local thrombolytics fail to resolve thrombi around catheters then they usually must be removed, sometimes with the institution of heparin and then warfarin for symptomatic relief.

The optimal therapy for upper extremity thrombosis either not associated with central venous catheters or after catheter removal is unknown and has not been studied in clinical trials. Extrapolating from therapy used at other sites of thrombosis, some authors recommend 5-7 days of heparin therapy with the institution of warfarin on the second day of heparin therapy.

Vena caval filters have traditionally been used when patients have contraindications to anticoagulation (such as active bleeding). As interventional radiologists became more proficient at filter placement in the 1980s, many began to wonder whether IVC filters might be used as primary therapy for deep venous thrombosis in patients with malignancies. A number of studies^5,21,25 have been done to address this issue but most of them are in the form of small case series and therefore IVC filter placement as primary therapy remains highly controversial. Most of the case series studies were done in patients with advanced malignancies (stage III or IV) and even in those studies where the safety and efficacy of IVC filters was demonstrated, patients had a very high mortality rate. Cohen et al.⁵, while demonstrating the safety of the procedure itself and concluding that filter placement was an effective approach to primary treatment, revealed that 56% of the patient population had died at a mean follow up period of 6.4 months. Another study²⁵ has shown an in-hospital mortality as high as 26% in those patients with malignancy who had filter placement for various different reasons. Other studies have consistently yielded similar numbers.³⁰

Only two studies have directly compared anticoagulation with IVC filter placement as primary therapy. The first⁴ was a retrospective study with a very small number of patients. In this study, both the anticoagulation group and the IVC filter group had similar high complication rates. However, the bleeding rate (35%) in the anticoagulation group is quite out of line with that demonstrated in other studies with more patients.³ It is difficult to draw any serious conclusions from this study due to the small number of patients included and its excessively high hemorrhage rate.

The second study²⁶ is a very complicated decision and cost-effectiveness analysis comparing observation, anticoagulation and IVC filter placement as primary therapy for deep venous thrombosis. This study draws on an incredibly large pool of studies and makes sweeping judgments by assigning "quality of life factors" (by distributing a graduated point system for different complications that may arise from different forms of therapy). This study concludes that anticoagulation and vena caval filter yield similar quality of life outcomes but that vena cava filter insertion is more cost-effective than anticoagulation. Observation alone was far inferior to both other strategies. This study is very difficult to interpret because it bases its conclusions on the cumulative data of many different studies. Additionally, a panel of "experts" designs a point system to rate the severity of different complications in order to determine quality of life issues.

Until a randomized prospective study is performed comparing anticoagulation with primary therapy with IVC filter placement, anticoagulation should probably be considered the standard of care when not contraindicated.

DVT as a Marker For Occult Malignancy: Since patients with cancer have clearly been shown to be at high risk for developing deep venous thrombosis, more recent research has been conducted in an attempt to determine whether the development of a deep venous thrombosis, particularly in patients with no other risk factors for DVT, may serve as a marker for a previously undiagnosed malignancy.

Gore et al.¹¹, in an early study, had founded an association between occult malignancy and deep venous thrombosis. In that study, 128 patients who had angiographic evidence of pulmonary embolism were retrospectively compared to 128 patients who had been suspected to have a pulmonary embolism but had negative pulmonary angiograms, thus ruling out embolism. The incidence of documented cancer prior to pulmonary angiography was similar in both groups. However, in the two years after angiography, cancer was subsequently diagnosed in 14% (n=13) of patients with angiographic evidence of embolism as opposed to no patients in the control group. This study was severely limited by the fact that follow up was limited to review of the medical record and was very incomplete. A total of 38 patients in the study group had either incomplete follow up or were completely lost to follow up. The number with either incomplete or no follow up in the control group was 50—to some extent the follow up was so poor that this study may even be dismissed. However, it did lay the groundwork for studies that would follow.

Aderka et al.¹ retrospectively compared a group of 35 otherwise healthy patients with "idiopathic’ DVTs with a control group of 48 patients diagnosed with DVTs but felt to have a clearly recognized risk factor for DVT (such as trauma, post-operative congestive heart failure, prolonged immobilization, polycythemia, postpartum, etc.). Patients in the idiopathic group had no recognizable predisposing risk factors for deep venous thrombosis. In an average follow up period of 44.9 months, 12 (34%) of the patients in the "idiopathic" group versus two (4%) of the patients in the control group were subsequently diagnosed with malignancy (p=.001). The origin of the malignancies detected the earliest after the diagnosis of venous thrombosis included the female reproductive organs and the male prostate. A subgroup analysis, comparing those patients in the idiopathic group who were subsequently diagnosed with malignancy with those in the same group who did not develop cancer, revealed that patients who were later diagnosed with cancer: 1) were significantly older (p<.01), 2) had a significantly higher percentage of eosinophils at presentation with DVT (p<.01), and 3) had a lower hemoglobin concentration at presentation (p<.02) than those who were not subsequently diagnosed with malignancy in the follow up period. Subsequent studies have not borne out this relationship with age and, as we shall see, have in fact suggested the opposite. The study detailed above has several major flaws in that, because of its retrospective nature, there was no standardized history and physical obtained at the time of admission for venous thrombosis. There is no clear indication of the quality of the work up of each patient who presented with a DVT. Could more malignancies have been diagnosed initially if a more careful initial screening of patients with "idiopathic" DVT had been performed? Also, the diagnosis of deep venous thrombosis was obtained in many different ways including venography, ventilation perfusion scanning, and Doppler examination of the legs and therefore we cannot be sure that the diagnosis of DVT was established equally well between the comparison groups. Lastly, the subsequent development of malignancy in 34% of patients with "idiopathic" DVT is quite a bit higher than that derived from other studies that we will review.

Goldberg et al.⁹ took a different approach to the problem. In this prospective, nonconcurrent, epidemiologic study all patients diagnosed with deep venous thrombosis by impedance plethysmography in a five year period were compared to a group of patients who had clinically suspected DVTs but negative IPG tests. Of 1546 patients tested by IPG, 103 patients were excluded because of a previous diagnosis of cancer. Of the remaining patients, 370 were diagnosed with DVT by IPG. Follow up of all patients was determined by chart review, mailed questionnaires, and death certificate searches. Surprisingly the authors were able to achieve a 100% follow up rate in those diagnosed with DVT and a 96% follow up rate in those who had been ruled out for DVT with an average follow up of 30.1 months and 34.7 months respectively. As determined by a life table approach, 6.3% (n=22) of patients diagnosed with DVT compared to 2.4% of patients without DVT were subsequently diagnosed with malignancy (p<.001). The relative risk for the discovery of a malignancy in patients diagnosed with a DVT by IPG was 2.7 (1.5-4.7) in the group studied. A subgroup analysis by age showed that the relationship between the diagnosis of DVT and the subsequent discovery of malignancy was not statistically significant for any one age group accept for those patients under 50 years of age. Patients under 50 years old were found to have a relative risk of 19.0 (2.2-168) when compared to those without DVT in the same age group. This is opposed to the observations of Aderka¹ noted above but more in line with a more recent study²² which we will discuss below. This study is somewhat different in that it is an epidemiologic study that compares all comers with DVT with a control group consisting of patients without DVTs. There is no attempt to distinguish idiopathic DVTs from DVTs secondary to known risk factors. While this study is fraught with difficulties in interpretation because it relies on chart review and questionnaires to analyze for a specific outcome, it would seem to suggest that being diagnosed with a DVT, regardless of extenuating circumstances occurring around the time of its development, places one in a higher risk category for being diagnosed with a subsequent malignancy—especially in those patients under fifty years of age.

The best study of its kind which seems to provide strong evidence that deep venous thrombosis may serve as a harbinger for occult malignancy was done more recently by Prandoni et al.²² in 1992. This was a prospective, cohort study in which 205 consecutive patients diagnosed with deep venous thrombosis by venography were divided into two groups—those with primary or idiopathic DVT (n=145) and those with DVT thought to be secondary to a well-described risk factor (n=105). This division of patients into an "idiopathic" study group and a "secondary" cohort group is similar to that of Aderka¹ which we discussed above. All patients in this group had a thorough history taken at the time of diagnosis with DVT as well as a routine physical, including pelvic, rectal, and breast examinations. Additionally, at the time of presentation with deep venous thrombosis, measurements of the erythrocyte sedimentation rate, complete blood count, liver and renal function tests, urinalysis, and chest radiographs were obtained. Patients were seen in follow up three months after presentation with a DVT and then every six months thereafter at which time a standardized history and physical was performed with suspicious results leading to further work up. The physicians who performed follow up exams were unaware of the classification of the type of DVT (i.e. idiopathic vs. secondary). Follow up was continued for two years.

At the time of diagnosis of DVT, 76 patients were felt to have clinical findings suspicious enough for malignancy to warrant further work up. Five of these patients were found to have malignancy after more extensive work up—all of these were classified in the idiopathic group initially. These five were excluded from further follow up and were not included in final data analysis. Five other patients were excluded who could not be followed up leaving a total of 250 people included in the study as noted above. During follow up, symptomatic malignant disease developed in 11 patients (7.6%) in the idiopathic group and in two (1.9%) in the secondary group (p=.043) for an odds ratio of 2.3 (1.0 to 5.2). Interestingly, of patients in the idiopathic group diagnosed with recurrent deep venous thrombosis (n=35) during the course of the study, 17.1% were subsequently diagnosed with overt malignancy. This incidence was quite statistically significant when compared to both the incidence of the development of malignancy in the secondary venous thrombosis group, with an odds ratio of 9.8 (1.8 to 52.2), and to the incidence of the development of overt malignancy in the idiopathic group, with an odds ratio of 4.3 (1.2 to 15.3).

As in the Goldberg⁹ study, deep venous thrombosis preceded the development of clinically overt carcinoma more commonly in younger patient population. Fifty percent of the malignancies detected in this study in the idiopathic group were in the thirty percent of patients under sixty years old. Because of its prospective nature, this study provides strong evidence that the development of idiopathic deep venous thrombosis may serve as a marker for a clinically undetected malignancy. The authors conclude that since many of the cancers became clinically apparent within the first year of the study, it is likely that the malignancies were present but clinically quiescent aside from their effect on the coagulation system at the time of enrollment in the study.

Summary: The implications of these studies are, however, unclear. It is unknown whether an extensive work up is indicated at the presentation with an idiopathic DVT. For an extensive malignancy work up to be warranted, it must be shown: 1) that clinically silent malignancies may be detected by extensive diagnostic testing, 2) that the discovery of malignancies by extensive testing at the time of presentation with deep venous thrombosis results in clinically significant improvements in the primary outcomes of morbidity and mortality, and 3) that the cost of extensive testing in terms of its financial burden to society as well as its potential morbidity and discomfort to the patients involved are outweighed by potential benefits to those that may be helped by the early detection of malignancy. These questions remain unanswered because no randomized controlled trials have compared extensive testing at the time of presentation with DVT to less aggressive diagnostic approaches.

There is rather strong evidence that suggests that the development of an idiopathic deep venous thrombosis may serve as a marker for clinically undetected malignancy, particularly in patients under sixty years of age. There is also strong evidence to suggest that the development of recurrent idiopathic deep venous thrombosis places one in a high risk category for subsequently developing an overt malignancy that may have been silent at the time of presentation with DVT. An extensive diagnostic work up at the time of development of an idiopathic DVT cannot be recommended at this time due to the fact that no randomized studies have been performed to evaluate this approach. For the time being, it is recommended that all patients who present with deep venous thrombosis be evaluated by a thorough history and physical exam which should probably include rectal, breast, and pelvic examinations. Further work up should then be guided by abnormalities detected during this initial history and physical.