Screening for cancer in venous thromboembolic disease

The association between cancer and venous thrombosis has long been recognised.¹,² Though venous thrombosis is a well known complication of established cancer, it might also be a marker of an otherwise occult cancer. If so, this raises the issue of whether otherwise healthy patients presenting with a venous thromboembolic event should be investigated for a possible underlying cancer on the grounds that a cancer diagnosed early may be more amenable to cure.

Whether screening for an underlying cancer is a good use of resources will depend on how common such cancers are; the cost, accuracy, and acceptability of the screening tests; and, most important, whether early detection of such cancers would improve patient outcome.

Large prospective studies of patients presenting with venous thromboembolic disease, linking hospital records with national cancer registers, find an incidence of previously undiagnosed cancer of 4-6.5%, giving standardised incidence ratios of 1.3-3.2.^3–5 Smaller retrospective and prospective studies looking particularly at patients with no obvious risk factors for their thrombosis find higher incidences of cancer, of 7.3-12% compared with 1.9-2.9% for patients with risk factors.^6–8 In these studies patients were not specifically investigated for an underlying cancer, the diagnosis being made after routine investigation on admission or after 6-12 months' follow up. Two studies in which patients underwent intensive investigations for cancer at the time of presentation found an incidence as high as 19% in patients with no risk factors.⁹,¹⁰

What investigations might prove useful as screening tests for occult cancers? When routine investigations were supplemented with abdominal and pelvic computed tomography or ultrasound scanning and carcinoembryonic antigen measurements in 424 patients with suspected venous thrombosis⁹,¹⁰ cancer was diagnosed in 33 patients (7.7%). In 21 cases the diagnosis was made on the basis of history, examination, or routine tests, but in the remaining 12 cases the diagnosis was made only after scanning or a carcinoembryonic antigen measurement, a detection rate of only 3.2%. This would have been appreciably higher if only patients with no risk factors had been so investigated.

In contrast, in a study of 136 patients with no risk factors for thromboembolism cancer was diagnosed in 16 patients (12%), all of whom had at least one abnormal finding on history, a thorough examination, full blood count, or chest x ray examination.⁷ Only 56 patients could have been classified as entirely normal, and in these patients no cancers were found at the time of presentation or on subsequent follow up. In this study a potential 59% of patients would have required further investigations for cancer, with at best a detection rate of 20%. In a second study of 326 patients, 10 of 13 cancers were diagnosed on the basis of an abnormal history, examination, full blood count, liver function tests, or urea and electrolyte measurements.⁸

Would early detection of these cancers improve patient outcome? Possibly, if the patient has a carcinoma of the breast, ovary, colon, or cervix, but there is no evidence for improved outcomes in carcinomas of the lung, brain, prostate, or pancreas, all of which have been associated with venous thrombosis. Furthermore, we cannot assume that these apparently occult cancers are indeed at an early stage of their development since they have already had a major clinical impact.

One study has recently reported on the survival of patients who were diagnosed with cancer at or around the time of presentation with a thromboembolic event.¹¹ When these patients were compared with age matched controls, with similar cancers but without an associated thrombosis, 44% were found to have metastases at the time of diagnosis compared with 35% of controls. One year survival was only 12% compared with 36% in the controls. If the cancer was diagnosed within 12 months of a thromboembolic event one year survival was 38% compared with 47% in the controls. The study concludes that patients with cancer diagnosed at or around the time of a thromboembolic event have a significantly worse prognosis than those patients without such an association.

So would screening for cancer in patients who present with a venous thromboembolic event be an effective use of resources? In the absence of an obvious risk factor for thrombosis there is a clear increase in the incidence of an underlying carcinoma in these patients. Estimates range from 7.3% to 19% at the time of presentation. Assuming an incidence of perhaps around 10%, screening for cancer becomes a reasonable option. On the basis of current evidence, however, intensive investigation cannot be recommended.

Firstly, cancers associated with venous thrombosis seem to have a relatively poor prognosis, and early diagnosis of many of these cancers has not been shown to improve survival. Secondly, we should not underestimate the potential harm to patients, both psychological and physical, associated with any kind of screening programme, as increasingly invasive investigations may be used to follow up abnormal screening tests for what may turn out to be benign or untreatable disease.

A practical approach would be always to consider the possibility of an underlying cancer in patients presenting with a venous thrombosis, particularly if they have no underlying risk factor for the thrombosis. We should take a careful history and make a thorough examination and do the simple routine blood tests—full blood count, liver function tests, urea and electrolytes—and a chest radiograph. This simple screen will lead us to the diagnosis in most patients with an underlying cancer. Further investigations would depend on the results of these tests. Before recommending more intensive investigations we need the results of a large randomised prospective study to assess whether incorporating investigations such as tumour markers, faecal occult blood, and computed tomography into a screening protocol will lead to improved outcome for those patients found to have an underlying cancer.