Elizabeth J Westgate Institute for Translational Medicine and Therapeutics, University of Pennsylvania , Garret A FitzGerald Institute for Translational Medicine and Therapeutics, University of Pennsylvania

Originally printed in:
» European Respiratory Disease 2006 - October 2006

Presentation of Case

A 25-year-old woman presented with pulmonary oembolism. She had been taking, without apparent complication, norgestimate/ethynil estradiol (Ortho Tri-Cyclen; 0.180, 0.215, and 0.250mg norgestimate cycles and 35μg ethinyl oestradiol) for two years, followed by the same drug combination but with a lower dose of ethinyl oestradiol (Ortho Tri-Cyclen Lo; 25μg ethinyl estradiol) for 14 months. A nonsmoker, she lacked a relevant family history and was vigorously athletic.

One month prior to presentation, she developed neck pain; disc protrusion at C5–C6 was detected by magnetic resonance imaging (MRI). The patient was prescribed valdecoxib, 20mg twice a day (bid), for two weeks. Her neck pain resolved. However, towards the end of this treatment period, she developed left-sided pleuritic chest and shoulder pain after a six-hour car ride. She was started on cyclobenzaprine, 10mg bid, and continued on valdecoxib. Her left-sided pain abated gradually. However, 18 days later, she developed right-sided chest and shoulder pain. A diagnosis of left iliac vein thrombosis and bilateral pulmonary emboli was based on a computed tomography (CT) scan. She was heparinised and continued on therapy with warfarin, 5mg bid, and enoxaparin, 60mg bid. Despite this, a ventilation-perfusion scan performed 13 days later showed multiple pulmonary emboli.

Discussion

Coxibs, selective inhibitors of cyclooxygenase-2 (COX-2), increase the risk of myocardial infarction and stroke,^1-3 prompting concern for patients with established cardiovascular disease. Caution may also extend to individuals predisposed to thrombosis by genetic or environmental factors.

Risk factors for spontaneous thrombosis include oral contraceptives, genetic predisposition to a hypercoaguable state, and prolonged periods of stasis.^4-6 At least two risk factors pertained to this patient.

The patient had been taking oral contraceptives for three-year prior to the index event, albeit without recognised thrombotic complication. It is possible that, despite this extended period of apparent tolerance, the embolic events were solely related to use of the oral contraceptives.⁴

The relatively small risk of venous thromboembolism attributable to oral contraceptive use may interact geometrically with the similarly small absolute risk of a procoagulant mutation, such as Factor V Leiden.⁵ However, documented genetic risk factors, such as abnormalities in lupus anticoagulant, anti-thrombin III, proteins C and S, plasma homocysteine, anticardiolipin, μ2 glycoprotein antibody and prothrombotic mutations in Factor V and prothrombin, were excluded. It remains possible that the patient was genetically predisposed to thrombosis by a mutation in an undetermined factor.

Finally, prolonged stasis, such as that which occurred during the six-hour car trip, may account for the clinical event.⁶ However, the absolute risk is small, and the patient had made this trip on multiple occasions devoid of apparent clinical complications during the prior three years.

The risk of thrombosis from valdecoxib has now been established,^3,7 and this, together with its propensity rarely to cause Stevens-Johnson syndrome without a mitigating benefit over traditional nonsteroidal antiinflammatory drugs, has led to its withdrawal from the market. The cardiovascular hazard of coxibs appears likely to be attributable to the suppression of COX-2- derived prostacyclin.^1,8 Deletion of the prostacyclin receptor in mice does not cause spontaneous thrombosis, but rather enhances the response to thrombotic stimuli.⁹ This is consistent with the fact that a cardiovascular signal from a coxib is most easily detected in patients with haemostatic activation, such as was observed under placebo-controlled conditions in two trials in patients undergoing cardiopulmonary bypass grafting7 and anecdotally in patients with connective tissue disease.¹⁰

Although this case does not establish a causative linkage with valdecoxib, the clinical event was not manifest until three potential risk factors were combined (oral contraceptives, prolonged stasis, and coxib treatment). Given the multiplicative interactions of risk factors for thromboembolic disease and the apparently untoward concurrence of two of them – the contraceptive pill and frequent road trips for the preceding three years – the rapid occurrence of the clinical event following initiation of the coxib suggests a causative link to the COX-2 inhibitor. However, it also remains formally possible that this temporal relationship was a coincidence.

This case report has been reported to the regulatory authorities and to the manufacturer of valdecoxib.

Just as the small absolute risks of thrombosis attributable to oral contraceptives and prothrombotic mutations may interact dramatically,⁵ selective inhibitors of COX-2 may also interact with genetic and environmental factors that predispose to the risk of thrombosis.

This article was first published in the Public Library of Science (PLoS) Medicine (2005);2(7): p. e197.