Unsuspected pulmonary embolism on CT scanning: yet another headache for clinicians?

Short abstract

Arguments for and against treatment of small unsuspected pulmonary emboli

Clinicians have long known that the symptoms and signs of pulmonary embolism are non‐specific and that, unless the index of suspicion is reasonably high, the diagnosis is frequently overlooked. Against this background, clinically unsuspected pulmonary emboli are increasingly being spotted by radiologists on CT scans. Clinicians not only need to be aware of this, but also need to know how to deal with such serendipity. The quality of CT examinations has improved unimaginably; image acquisition, particularly with the new generations of multidetector CT (MDCT) machines, is now astonishingly fast and access to CT scanning has increased. The entire thorax can now be covered in a single breath‐hold, and image degradation due to respiratory and cardiac motion is no longer a major issue. Furthermore, because of narrow collimation, images with exquisite spatial resolution are almost the norm and visualisation of opacified peripheral pulmonary arteries (down to fifth order branches) is now possible.¹,²

The detection of unsuspected pulmonary emboli by CT scanning has been the subject of a number of previous reports.³,⁴,⁵,⁶,⁷,⁸ The prevalence of incidental emboli in these studies has varied from 0.6% to 5% and has depended to a large degree on the tests used for detection (single‐slice CT vs MDCT vs echocardiography), the manner in which images were reviewed (hard copy vs workstation analysis) and the demographics of the study population (cancer vs non‐cancer or inpatient vs outpatient).³,⁴,⁵,⁶,⁷,⁸,⁹

In the current issue of Thorax (see p 536), another study documents the prevalence of incidental pulmonary emboli in consecutive inpatients undergoing MDCT scanning.¹⁰ Most patients were imaged on a 16‐channel CT machine and, following the routine report, all studies were reviewed by an experienced thoracic radiologist unaware of the original findings. In nine out of 28 studies judged positive by the expert reviewer, filling defects were overlooked by the initial reporting radiologist but, in these patients, emboli were localised to the segmental (n = 6) and subsegmental (n = 3) vessels. Overall, emboli unsuspected at the time of referral were found in 28 of 487 (5.7%) scans. This figure is broadly comparable to the inpatient prevalence of unsuspected emboli in previous CT studies: in two reports comprising 634 inpatients the prevalence of unsuspected filling defects was 4.3%.⁷,⁸ An interesting aspect of the present study, implied but not stressed in earlier series, was the significant relationship between the likelihood of unsuspected pulmonary embolism and age. Unsuspected emboli were not detected in any of the patients aged below 50 years (n = 47), whereas incidental filling defects were present in almost 17% of those aged over 80 years (n = 66). The authors suggest that age may have been a surrogate for other relevant risk factors. However, an alternative explanation is that pulmonary emboli might simply be missed in the elderly because of the tendency to ascribe symptoms to age‐related comorbidity.

The links between cancer and venous thromboembolism are widely acknowledged and one theme of earlier publications on incidental pulmonary emboli has been the emphasis on a background of malignancy. In the studies reported by Gosselin et al⁷ and Storto et al⁸ there was a history of malignant disease in 70% and 83% of patients, respectively. At first sight these results seem to imply that radiologists should be on the lookout for clinically unsuspected pulmonary emboli in patients with cancer. However, the published data warrant closer scrutiny for there is an intrinsic bias: patients with cancer are just more likely than those without to have a CT examination because of the need for pretreatment staging, to say nothing of the further surveillance CT studies needed to monitor treatment response or detect relapse. It is therefore perhaps unsurprising that patients with malignancy account for a larger proportion of those with unsuspected pulmonary embolism on CT scans. It is noteworthy that clinically unsuspected emboli were seen in fewer than 10% of the 81 patients with cancer in the study by Gosselin et al.⁷ The experience at tertiary cancer centres is similar: the prevalence of non‐suspected pulmonary embolism was only 4% (16/403 patients) in a recent study published from the M D Anderson hospital.¹¹ Importantly, the results presented by Ritchie and colleagues show, for the first time, that there are no significant differences in the prevalence of incidental pulmonary embolism between patients with and without malignancy (18/343 (5.2%) vs 10/144 (6.9%)). Thus, two points emerge: (1) despite the high prevalence of malignancy in patients with incidental pulmonary embolism, the converse scenario (that patients with cancer are more likely to have unsuspected pulmonary emboli) does not necessarily follow; and (2) the demonstration of incidental emboli should not, in the absence of some clinical suspicion or other evidence, lead to a barrage of radiological tests in the forlorn hunt for an underlying neoplasm.

The above notwithstanding, there is perhaps a more pressing clinical issue. Indeed, most radiologists would readily admit that the task of detecting and reporting unexpected emboli is not particularly taxing. In contrast, once pulmonary embolism is reported, the clinician must make decisions on management and these are seldom easy. The pragmatist would argue that, unless there are contraindications, anticoagulation is warranted once a diagnosis of pulmonary embolism is made, incidental or otherwise. This certainly appears to be the case at the institution from which Ritchie and colleagues report their data¹⁰ and probably mirrors clinical practice elsewhere.¹² The presumed logic is that, because the CT examination is a “snapshot”, the finding of an unsuspected filling defect (even when this is small or isolated to a segmental or subsegmental branch) might be the harbinger of trouble. Deviating from this mindset is uncomfortable because little is known about the natural history of untreated pulmonary embolism. Indeed, as Goodman¹³ has stated so cogently, anticoagulation therapy for acute pulmonary embolism became a part of medical culture so rapidly following the publication of the results of the trial by Barritt and Jordan¹⁴ and at a time when small filling defects were almost certainly overlooked by the tests then available, that it is now difficult to know what to do with small and incidentally diagnosed emboli.

In the absence of convincing evidence from controlled studies, the prognostic significance of and treatment implications for incidental pulmonary emboli are not known. However, there are indirect data which might guide management in some patients. First, it is known that not all pulmonary emboli are fatal: post‐mortem studies indicate that between 9% and 63% of pulmonary emboli are incidental and probably do not contribute to death.¹⁵,¹⁶,¹⁷,¹⁸,¹⁹ Second, as Engelke and colleagues have shown recently (albeit in a retrospective fashion), patients with missed CT diagnoses of pulmonary embolism who are then not anticoagulated do not necessarily have an adverse outcome.²⁰ In line with this, Eyer et al¹² have also reported that, in 25 untreated patients with isolated subsegmental emboli, there were no deaths attributable to pulmonary embolism. Third, although their study was not designed to answer questions about clinical significance, Ritchie et al have shown that most incidentally diagnosed pulmonary emboli are small and located in segmental and subsegmental vessels; the clinical or prognostic significance of these small filling defects is far from clear.²¹ Fourth, there are grounds for believing that one of the functions of the lung is to filter small (and possibly physiological) emboli;²² the obvious question is whether, because of the excellent anatomical resolution of MDCT, we are only now identifying these apparently “innocent” emboli? Finally, it is perhaps telling that, although more peripheral emboli were seen on MDCT scanning than with a single‐slice scanner, there was little difference in clinical outcome in one study.²³ Presumably based on similar data and as practical as ever, Goodman has gone so far as to argue that, when the risks of anticoagulation are unacceptable, treatment for incidentally diagnosed emboli might justifiably be withheld if cardiopulmonary reserve is satisfactory, the filling defects are small and there is no evidence of deep venous thrombosis.¹³

It will be clear to the reader that, as technology evolves, radiologists will increasingly report chance findings on CT scans. The onus is on clinicians to decide what is and what is not clinically important. This is not a new problem: the obvious drawback of super‐sensitive tests such as MDCT is that “abnormalities” which ultimately have no clinical impact may be identified. On a philosophical note, there are intriguing parallels outside the sphere of pulmonary embolic disease. For instance, as radiologists who regularly report thoracic CT scans will testify, small incidental pulmonary nodules are all too common on MDCT. Another area in which the detection of clinically questionable change arises is in connective tissue disorders where, because there is an expectation of lung disease, patients are often screened with CT. However, in this scenario, a problem that clinicians often face is what to do when—as occurs not infrequently—limited and physiologically “silent” disease is reported by the radiologist.

Whatever the arguments for and against treatment of small unsuspected pulmonary emboli, it is hoped that data such as those presented by Ritchie et al will fuel pragmatic prospective studies. After all, it does seem remarkable now that a single paper¹⁴ published over 45 years ago (and questioned because of concerns about potential confounding factors²⁴) has essentially dictated the treatment of pulmonary thromboembolism. Although it seems unlikely that ethics committees would approve a placebo‐controlled trial of treatment in symptomatic pulmonary embolism, the same reservations may not apply to patients with clinically unsuspected and haemodynamically insignificant pulmonary thromboembolism. The time to address these hitherto unanswered questions may be upon us!