Since the quality of computed tomography (CT) imaging techniques has improved, pulmonary embolism (PE) has increasingly been detected incidentally on routine CT examination. In particular, in patients with malignancy, who are known to be at elevated risk of developing venous thromboembolic events, and who frequently undergo CT scanning for reasons such as diagnosing, staging, and treatment evaluation, incidental PE has become a relatively common finding .
There is limited information on the embolic burden in patients in whom PE goes clinically unnoticed. It may be theorized that, in patients without symptoms suggestive of PE, the emboli may more frequently be localized distally, leading to a smaller obstruction index.
The first aim of the present study was to assess the embolic burden of patients with incidental PE and compare it with that of patients with symptomatic PE. The second aim was to assess the impact of the embolic burden on the outcome of the patients.
Consecutive adult patients with active malignancy who received a diagnosis of incidental PE between January 2003 and July 2012 in our hospital were included. The characteristics of this cohort have, in part, been described previously . Incidental PE was defined as a diagnosis of PE detected on CT scans ordered for reasons other than suspected PE . Institutional review board approval was waived for this observational and retrospective study.
Imaging was performed with multidetector CT scanners (four-slice, 16-slice, 64-slice and 320-slice CT scanners; Toshiba, Otawara, Japan). Images were reconstructed with a slice thickness of 1.0 mm. All CT images were reviewed on a PACS workstation by an experienced thoracic radiologist (L.J.M.K.), who was blinded to the original CT report, the location of the filling defect, and the clinical information of the patients. Both the degree of pulmonary artery obstruction, according to the scoring system of Qanadli, and the largest pulmonary artery involved, i.e. central or interlobar, segmental, or subsegmental, were assessed. The Qanadli obstruction index was defined as the number of segmental artery branches that are blocked, with one point assigned for partial blockage, or two points for complete obstructive PE. With this scoring system, 40 is the highest possible score, corresponding to a 100% obstruction index . To compare the obstruction indexes of patients with incidental PE with those of patients with symptomatic PE, we used a previously described cohort of 113 consecutive patients with acute symptomatic PE diagnosed on CT pulmonary angiography (CTPA) as a reference group [2, 5].
A retrospective chart review was performed to record the clinical outcome. All patients were followed for 6 months for the occurrence of death. The cause of mortality was assessed by reviewing the pathology report. In the cases in which an autopsy was not performed, the likely cause of death was verified with the treating physician by reviewing the medical records and death certificates.
The obstruction indexes of patients with incidental and symptomatic PE were compared for statistical difference with the Mann–Whitney test. Cox regression analyses were performed to assess the impact of the level of obstruction on survival during 6 months of follow-up. The hazard ratio (HR) was adjusted for potential confounders, including age, gender, and stage of the malignancy (i.e. localized vs. metastatic). spss version 20 (SPSS, Chicago, IL, USA) was used for all analyses.
During the study period, incidental PE was diagnosed in a total of 65 patients. The original CT images could not be retrieved for three patients, and these were excluded from the present analyses. In 13 patients, incidental PE was diagnosed on an abdominal CT without a complete chest CT examination, and these patients were also excluded from the present analysis assessing the embolic burden. One other patient was excluded because expert opinion refuted the definite presence of PE. The characteristics of the remaining 48 patients are shown in Table 1.
|Incidental PE (n = 48)||Symptomatic PE (n = 113)||P-value|
|Age in years (mean ± SD)||63 ± 15||55 ± 17||0.01|
|Male sex, n (%)||27 (56.3)||60 (53.1)||0.52|
|Inpatient at time of diagnosis, n (%)||12 (25.0)||20 (17.7)||0.20|
|Recent immobilization, n (%)||17 (35.4)||20 (17.7)||0.01|
|Recent surgery, n (%)||3 (6.2)||22 (19.5)||0.025|
|Obstruction index, median % (IQR)||18 (10–30)||30 (10–53)||0.008|
Expert reading identified 19 (39.6%) patients with central PE, 29 (60.4%) with segmental PE, and no patients with subsegmental PE. The median pulmonary obstruction index was 18% (interquartile range [IQR] 10–30%). No correlations were found between obstruction index and age, sex, and type or stage (localized or metastatic disease) of the malignancy. In the cohort of patients with symptomatic PE diagnosed with CTPA, the median obstruction index was 30% (IQR 10–53%), which was significantly higher than that of the patients with incidental PE (P = 0.008).
During follow-up, 16 patients (33.3%) died. Of these, 13 died because of progressive malignant disease, two patients experienced fatal bleeding, and one died because of an infection. The median obstruction index was similar for those who survived and those who died during the first 6 months following PE diagnosis (17.5 vs. 21.3%, P = 0.68). On Cox regression analysis, no association was found between the obstruction index and all-cause mortality during follow-up: adjusted HR 1.006 (95% confidence interval 0.97–1.04).
The present study is the first to systematically investigate the location and extent of incidental PE diagnosed on CT scans that were conducted for reasons other than the identification of PE. An important observation was that, in 40% of the patients, incidental PE was localized in the central pulmonary arteries, which is comparable to the proportion of centrally located emboli observed in patients with symptomatic PE . This highlights the fact that even central PE may go clinically unnoticed. The pulmonary artery obstruction index in patients with incidental PE was lower than that in the reference sample of symptomatic PE patients diagnosed with CTPA; however, it should be noted that the image quality for the detection of PE on the staging scans was possibly insufficient to detect all PE filling defects, and this may have led to an underestimation of the obstruction index. The obstruction index reported should thus be regarded to be as least as high as observed; dedicated CTPA investigations would possibly have identified more emboli. The obstruction index was not associated with mortality during follow-up.
Remarkably, isolated subsegmental PE was not identified in any of the patients. This may imply that isolated subsegmental emboli cannot be detected well on cancer-staging CT. Again, this is probably because administration of contrast was not specifically timed for the identification of PE, and the image quality was therefore suboptimal for the detection of smaller, particularly subsegmental, emboli. Notably, even in symptomatic patients diagnosed with CTPA, the difficulties of diagnosing PE in subsegmental vessels have well been documented .
The limitations of this study include the fact that, in this retrospective study, we did not specifically assess for the presence of PE-related symptoms at the time of diagnosis. The clinical presentation of PE may sometimes be subtle and easily overlooked, or attributed to the underlying malignancy. In a retrospective case–control study, O'Connell et al.  found symptoms possibly caused by PE, including fatigue and shortness of breath, to be significantly more frequent in cancer patients diagnosed with unsuspected PE than in cancer patients without PE. Therefore, we cannot state that all PE patients included in this analysis were completely asymptomatic. Nonetheless, the suspicion of PE was not raised in any of the ordered CT scans. Finally, CT scans with different detectors were used across the study period, and this may have influenced the detection of PE.
In conclusion, the observed embolus load of incidental PE was substantial, with centrally localized PE being seen in ~ 40% of the patients. However, the embolus load of patients with incidental PE was significantly lower than that of those with symptomatic PE. No association was found between the embolic burden and the occurrence of mortality during follow-up.