If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Radiology and Functional Imaging, Karolinska University Hospital, SwedenDepartment of Physiology and Pharmacology, Karolinska Institutet, Sweden
Over 9000 CT studies were reviewed, partly assisted by a deep learning algorithm.
•
In patients with cancer, unreported incidental pulmonary embolism was common.
•
Patients who were untreated had a high risk of recurrent venous thromboembolism.
•
A high proportion of the initial pulmonary embolism resolved without treatment.
•
Even in resolved initial pulmonary emboli the recurrent thromboembolic risk was high.
Abstract
Purpose
To assess the risk of recurrent venous thromboembolism (VTE) and death in patients with unreported cancer-associated incidental pulmonary embolism (iPE).
Materials and methods
Matched cohort study on cancer patients with a CT study including the chest between 2014-01-01 and 2019-06-30. Studies were reviewed for unreported iPE, and cases were matched with controls without iPE. Cases and controls were followed for one year, with recurrent VTE and death as outcome events.
Results
Of the included 2960 patients, 171 patients had unreported and untreated iPE. While controls had a one-year VTE risk of 8.2 events per 100 person-years, cases with a single subsegmental iPE had a recurrent VTE risk of 20.9 events, and between 52.0 and 72.0 events per 100 person-years for multiple subsegmental iPE and more proximal iPE. In multivariable analysis, multiple subsegmental and more proximal iPE were significantly associated with the risk of recurrent VTE, while single subsegmental iPE was not associated with the risk of recurrent VTE (p = 0.13). In the subgroup of patients (n = 47) with cancer not in the highest Khorana VTE risk category, no metastases and up to three involved vessels, recurrent VTE occurred in two patients (4.7 cases per 100 person-years). There were no significant associations between iPE burden and risk of death.
Conclusion
In cancer patients with unreported iPE, iPE burden was associated with the risk of recurrent VTE. However, having a single subsegmental iPE was not associated with the risk of recurrent VTE. There were no significant associations between iPE burden and risk of death.
Venous thromboembolism (VTE), which includes pulmonary embolism (PE) and deep venous thrombosis (DVT), is common in patients with cancer. A significant proportion of detected PE are unsuspected events [
], with the distinction that in subsegmental iPE, the recommendation is to offer treatment on a case-by-case basis considering potential benefits and risks of anticoagulation, particularly major bleeding. The clinical significance of subsegmental iPE is not completely understood. While a recent systematic review showed no difference in the risk of recurrent VTE and death between patients with treated and untreated subsegmental iPE [
], a study on cancer-associated subsegmental iPE and PE showed that the one year risks of recurrent VTE were 8 % for treated compared to 13 % for untreated patients [
Risk of recurrent venous thromboembolism and major hemorrhage in cancer-associated incidental pulmonary embolism among treated and untreated patients: a pooled analysis of 926 patients.
To evaluate the significance of subsegmental iPE in cancer patients, searching for previously unreported iPE can be considered. In previous retrospective studies between 32 and 75 % of iPE were unreported [
Incidental pulmonary embolism in patients with cancer: prevalence, underdiagnosis and evaluation of an AI algorithm for automatic detection of pulmonary embolism.
]. Evaluating the clinical outcome of patients with unreported and untreated iPE avoids the selection bias inherent in prospectively identified iPE, as actively abstaining from treatment could be based on an estimated low risk of recurrent VTE, or a high risk of recurrent VTE but with an associated unacceptably high risk of major bleeding.
The objectives of this study were to assess the risk of recurrent VTE and risk of death in patients with unreported and untreated iPE and to see if there were cancer-related factors and/or factors regarding initial iPE burden that were associated with adverse outcomes. A secondary objective was to evaluate the radiologic natural history of unreported iPE.
2. Material and methods
2.1 Design
A retrospective single-center matched cohort study conducted in Halland Hospital Halmstad, Region Halland, Sweden. The study protocol was approved by the Swedish Ethical Review Authority. Informed consent was waived because of the retrospective nature of the study.
2.2 Patients
Patients were identified by reviewing all CT requests in the Radiology Information System. Inclusion criteria were patients of at least 18 years old with a confirmed diagnosis of cancer, at least one contrast-enhanced CT study including the chest, and an indication for the study other than suspicion of PE or follow-up after reported PE, between 2014-01-01 and 2019-06-30.
Patients with unreported iPE after image review, identified as described below, were individually matched with patients (‘controls’) without iPE with respect to cancer type and stage, age (±five years), sex and date of the scan. Matching was done on a study-level to accommodate matching for date of the scan, and three studies were chosen for each patient with iPE. If a control had more than one matched study, only one study was chosen by random to determine the baseline date for collecting baseline variables and defining the follow-up period.
2.3 CT scan parameters
All studies were performed on a 64-slice multidetector CT scanner (Lightspeed VCT 64 or Revolution CT, GE Healthcare). Contrast media volume and injection rate were chosen using the Omnivis calculator (GE Healthcare). Contrast media volume and flow rate varied depending on patient characteristics and type of scan, with 375 mg iodine/kg for a chest CT and 500 mg iodine/kg for a combined CT of the chest and abdomen, up to a maximum dose corresponding to a body weight of 80 kg. For example, a 70-year-old male weighing 80 kg with normal renal function would receive 86 ml contrast media (Omnipaque 350 mg/ml) at a flow rate of 2.9 ml/s for a chest CT, and 114 ml at a flow rate of 3.8 ml/s for a combined CT scan of the chest and abdomen. Bolus tracking with monitoring of the attenuation in the descending aorta was used for all scans. For the combined protocols, dual acquisition was used. The scans were triggered when attenuation in the descending aorta reached >100 HU, with a fixed delay of 18 s for the chest scan and 63 s for the abdominal scan. In cases of an estimated glomerular filtration rate <45 ml/min, a reduced contrast media volume was considered on a case-by-case basis after reviewing the indication of the exam.
2.4 Image review
All included CT studies were reviewed for the presence of iPE, which was diagnosed if there were an intraluminal filling defect in one or more pulmonary arteries, not attributable to motion artifacts or inadequate contrast opacification of the arterial tree. The axial 0.625 mm slice thickness images were used for the retrospective review, with reconstruction in the coronal and sagittal plane as needed. The images were reviewed in the picture archiving and communication system (PACS; Sectra, Sectra AB). Four iPE groups were categorized: single subsegmental, multiple subsegmental, segmental, and lobar or more proximal iPE. In addition, the number of involved segmental and/or subsegmental vessels was estimated, where an embolus in a segmental or subsegmental artery was counted as one, while more proximal emboli were counted as the number of distally arising segmental arteries according to Qanadli et al. [
], but without weighting for degree of obstruction. As the assumption was that smaller iPE in a single vessel or a few vessels were more likely to be unreported, the number of vessels for each iPE case was emphasized over the arterial obstruction index. When an iPE was found, earlier studies were also reviewed to determine when the iPE first occurred. Date of first occurrence was used as the baseline date for collecting baseline data and determining the follow-up period.
To facilitate review, an artificial intelligence (AI) algorithm for automatic PE detection (Aidoc Briefcase, Aidoc Medical) was used for a subset of the studies. The AI algorithm is a commercially available deep learning cloud-based AI algorithm for PE detection and triage; in brief, it is a convolutional neural network trained and validated on tens of thousands of CT examinations acquired on a diverse range of CT scanners from multiple medical centers around the world.
The CT studies included were initially reviewed either by the first author who is a general radiologist with nine years of experience, the AI algorithm or both. For the AI analyses, all CT studies were pseudonymized, transferred via secure upload, and then analyzed by the AI algorithm. All studies marked as positive (suspicion of iPE) by the AI algorithm were reviewed again by the first author. The subset of studies reviewed by both the AI algorithm and the first author are reported in more detail in a previous study, with the AI algorithm having a sensitivity of 90.7 % and specificity of 99.8 % for iPE [
Incidental pulmonary embolism in patients with cancer: prevalence, underdiagnosis and evaluation of an AI algorithm for automatic detection of pulmonary embolism.
]. All included iPE cases were reviewed by the first author and the second author, a radiologist with six years of experience including one year of subspeciality training in thoracic radiology. Final iPE categorization were done by consensus.
2.5 Radiologic natural history of iPE
For patients with unreported iPE, subsequent CT studies were reviewed to follow the development of the original iPE and eventual new PE, until no PE was present or no further studies were available. Development was stratified at the first follow-up examination as complete regression, partial regression, no change, progression of the iPE, or new PE in different arteries.
2.6 Clinical data review and follow-up
A subset of baseline data was first extracted from the study requests (age, sex, cancer type and cancer stage) for the whole study population. For patients with iPE and the matched controls, the electronic health register (EHR) was used to confirm the cancer diagnosis and to record further baseline variables. Cancer types were grouped based on VTE risk according to Khorana et al.: very high (stomach, pancreas), high (lung, lymphoma, gynecologic, bladder, testicular) or normal risk of VTE [
All patients with iPE and all controls were followed one year from the baseline date. VTE was defined as an objectively confirmed DVT or PE, either diagnosed incidentally or by clinical suspicion, with information collected from the EHR and the PACS. Date of death was registered in patients who died during the follow-up period.
2.7 Statistics
Standard descriptive statistics were used to compare patients with unreported iPE and controls. Categorical data are presented as a percentage and continuous data as either the mean ± 1 standard deviation (SD) or the median with interquartile range (IQR). The chi square test was used for categorical variables, using Bonferroni correction for multiple comparisons. Kaplan-Meier curves and Poisson regression were used for univariate survival analyses, using the log-rank test and Wald test respectively to test for significance. Cox regression was used for univariate and multivariable analyses. Patients were censored at the end of follow-up, the date of the outcome event or if they moved out of Halland, and for recurrent VTE analyses: start of therapeutic-dose anticoagulants. A p-value <0.05 was considered significant. SPSS Statistics for Windows, version 27 (IBM Corporation) was used for all analyses.
3. Results
A total of 2960 patients were included, having undergone 9309 elective CT studies during the study time period. The inclusion/exclusion process is visualized in Fig. 1. In 350 studies (3.8 %) an iPE was found at the retrospective review, of which 252 (72 %) were unreported. There were no significant differences in the prevalence of iPE between the subsets of studies primarily reviewed by either the first author (152 iPE cases in 4347 studies: 3.5 %), AI (123 iPE cases in 3070 studies: 4.0 %) or reviewed by both (75 iPE cases in 1892 studies: 4.0 %). Patients with iPE were more likely to have a Khorana cancer type with high or very high risk of VTE, metastases, or being of female sex (Table 1).
Fig. 1Flowchart of the inclusion/exclusion process.
After exclusion for treatment dose anticoagulants at or within five days of the baseline date, 171 patients with iPE and 323 controls were included in the survival analyses. While there were no differences in age, sex or cancer type and stage between the cases and the matched controls, more cases than controls had active antitumoral treatment with either chemotherapy or targeted cancer therapy (71.9 % vs 57.0 %, p < 0.001).
3.1 Risk of recurrent VTE
During the follow-up period, 21 cases of VTE occurred in the controls (6.5 %) while 51 cases of recurrent VTE occurred in the cases (29.8 %). Of the controls, 81.0 % of VTE events were symptomatic while in the cases, symptomatic events ranged from 25.0 % to 50.0 % depending on iPE group, with a significantly higher proportion of symptomatic events in controls compared to cases with single subsegmental iPE (p = 0.022), multiple subsegmental iPE (p = 0.006) and segmental iPE (p = 0.005).
Univariate analysis of the risk of recurrent VTE stratified by iPE burden is presented in Fig. 2a , showing a significant association between iPE burden and recurrent VTE risk (p < 0.001). Of the control group, 21 were diagnosed with a VTE during follow-up (8.2 events per 100 person-years), while the VTE incidence rate varied from 20.9 events per 100 person-years for single subsegmental iPE to 72.0 events per 100 person-years for lobar or more proximal iPE, shown in Table 2. Fig. 2b shows the cumulative incidence of recurrent VTE with iPE burden stratified for number of involved vessels instead of the most proximal extent of the iPE. The recurrent VTE incidence rate were 26.6–30.6 events per 100 person-years for 1–3 involved vessels and 97.5–113.5 events per 100 person-years for 4 or more involved vessels. In multivariable analysis, Khorana cancer type, metastases and iPE burden were significantly associated with the risk of VTE (Table 3). Single subsegmental iPE was not significantly associated with the risk of recurrent VTE (p = 0.13).
Fig. 2a. Cumulative incidence of recurrent VTE in patients in different iPE groups, stratified for most proximal iPE level.
Post hoc analyses showed that in the 49 patients with iPE with up to three involved vessels (including patients with single subsegmental, 2–3 subsegmental, and 1–3 segmental iPE), no metastases and having a cancer type not in the highest Khorana VTE risk category, recurrent VTE occurred in two patients (4.7 cases per 100 person-years) during follow-up. In the 40 patients with up to two involved vessels recurrent VTE occurred in one patient, and in the 22 patients with a single involved vessel (subsegmental or segmental) no recurrent VTE occurred.
3.2 Risk of death
The risk of death stratified by iPE burden is presented in Fig. 3. In univariate and multivariable analyses, initial iPE burden was not associated with the risk of death (p > 0.05). In multivariable analyses, cancer type and metastases were significantly associated with the risk of death (Table 4). In cases, suffering a symptomatic recurrent VTE during follow-up was in univariate analysis associated with a higher risk of death compared to cases without a recurrent VTE (HR 2.49, p = 0.004), while in multivariable analysis this association was no longer significant (p = 0.60).
Fig. 3Cumulative incidence of death in patients in different iPE groups, stratified for most proximal iPE level.
After exclusion of patients without a follow-up CT study and patients starting therapeutic-dose anticoagulants before the first follow-up CT study, 121 patients were included for assessment of radiologic iPE development. Median time to the first follow-up CT study was 76 days (IQR 56–111 days). Of the 50 excluded patients 7 started therapeutic-dose anticoagulants due to diagnosis of a DVT, 22 died during early follow-up after a median of 48 days (IQR 29.75–86.25), and 21 did not perform a follow-up CT.
At the first follow-up CT, complete or partial resolution, without any new PE, had occurred in 86 patients (71.1 %). In 28 patients (23.1 %) there were progression of the original iPE or new PE. iPE development stratified by initial iPE burden is shown in Fig 4. In the 28 cases of progression or new PE, the original iPE showed total or partial resolution in 12 cases, no change in 3 cases, progression in 4 cases while in 9 cases the new PE were too extensive to allow assessment of the original iPE.
Fig. 4Radiologic natural history of the original iPE, in the different iPE groups.
In the group with partial or complete resolution of the initial iPE, 16 suffered a subsequent recurrent VTE (18.7 %) during the follow-up period.
4. Discussion
4.1 Risk of recurrent VTE
Unreported iPE were common in patients with cancer, with a total iPE prevalence of 3.8 %, of which 72 % were unreported. In patients with unreported iPE, iPE burden was associated with the risk of recurrent VTE, ranging from over 50 cases per 100 person-years for multiple subsegmental iPE to over 70 cases per 100 person-years for lobar or more proximal iPE. When stratifying instead for number of involved vessels and disregarding the most proximal level of the iPE, less than four involved vessels were associated with a lower risk of recurrent VTE (27–31 cases per 100 person-years) compared to four or more (98–114 cases per 100 person-years).
While having a single subsegmental iPE was not significantly associated with the risk of recurrent VTE, multiple subsegmental iPE had a clear increased risk of recurrent VTE. While the present study shows that increasing iPE burden is associated with an increasingly high risk of recurrent VTE, Kraaijpoel et al. and Van Der Hulle et al. showed that in 695 and 926 patients with cancer-associated iPE the risk of recurrent VTE was comparable between patients with subsegmental iPE and more proximal iPE [
Risk of recurrent venous thromboembolism and major hemorrhage in cancer-associated incidental pulmonary embolism among treated and untreated patients: a pooled analysis of 926 patients.
]. As the vast majority of patients received anticoagulants, their results likely reflect a comparable intrinsic VTE risk between patients with subsegmental and more proximal iPE, rather than a direct comparison in regard to the index event. While the ASCO guidelines recommend offering treatment on a case-by-case basis for subsegmental iPE, our results suggests that patients with multiple subsegmental iPE would benefit from treatment, while the risk-benefit ratio is still unclear concerning single subsegmental iPE. In addition, in the small subset of patients with a cancer type not in the highest Khorana VTE risk category, no metastases and up to three involved vessels recurrent VTE occurred in only two patients during follow-up, and in none of the patients with iPE involving a single vessel.
If a subsegmental iPE is to be left untreated, confirmation of the subsegmental location by an additional review can be considered as the agreement rate on the most proximal extent of a PE is moderate even between experts [
]. It is recommended to perform compression ultrasonography to exclude the presence of concurrent DVT, as earlier small studies have shown that up to 50 % of patients with iPE have a concurrent DVT of the lower extremities [
]. Another approach that can be considered in patients with cancer after exclusion of a concurrent DVT, is treatment with prophylactic dose anticoagulants. Supporting this approach is the fact that in the present study most iPE resolved spontaneously, and that few events of recurrent VTE occurred within the first weeks of follow-up. However, as the incidence of recurrent incidental VTE is influenced by the chosen scan interval for cancer monitoring, the short-term risk might be underestimated. Whether abstaining from treatment, or using a modified treatment protocol, is safe in patients with cancer is yet to be evaluated.
4.2 Radiologic natural history of iPE
The majority of the unreported and untreated iPE had completely or partially resolved at the first follow-up CT scan, including in patients with recurrent PE in different arteries. These results confirm earlier studies reporting on subgroups with untreated iPE. For example, Gladish et al. showed that in 12 patients with unreported and untreated iPE the initial emboli had in all patients resolved completely at the first follow-up, with new emboli in one patient [
]. However, even if the majority of the initial iPE had a benign natural history in the present study, cases with complete or partial resolution of the iPE still had a subsequent recurrent VTE risk of 18.7 %, showing that reliance on iPE resolution as a marker of lower recurrent VTE risk is not feasible.
4.3 Risk of death
Neither initial iPE burden nor suffering a recurrent VTE were associated with the risk of death.
4.4 Limitations
Standard CT examinations are not optimized for opacification of the pulmonary arteries, and even if the reported iPE prevalence is consistent with earlier studies the prevalence was likely slightly underestimated, as while 20 % of the included studies were reviewed by both the first author and the AI algorithm, 80 % were primarily reviewed by either the first author or the AI algorithm alone. In these subsets, after a negative first review no second review was performed.
The lower limbs were not examined for the presence or absence of DVT at baseline and how much of the increased recurrent VTE risk was conferred by a concurrent DVT cannot be evaluated.
No data were available on performance status or if there were symptoms attributable to the iPE at baseline. No data were available on the exact cause of death, precluding an analysis of PE-associated mortality.
5. Conclusion
In patients with unreported and untreated cancer-associated iPE, initial iPE burden was significantly associated with the risk of recurrent VTE. The risk was lower for patients with low initial embolic burden, but they still had a high rate of recurrent VTE, supporting current guidelines that cancer patients with iPE benefit from treatment, including most patients with subsegmental iPE. However, having a single subsegmental iPE was not significantly associated to the risk of recurrent VTE, and in the subgroup of patients with a Khorana cancer type with normal or high VTE risk, no metastases and up to three involved vessels recurrent VTE occurred in only two patients.
There were no significant associations between iPE burden at baseline and risk of death.
Further studies are needed to assess whether anticoagulation can be withheld in selected cancer patients, or whether treatment with prophylactic dose anticoagulants is a feasible alternative to standard anticoagulant regimens.
Funding
Peder Wiklund has received grants from the County Council of Halland and Sparbankstiftelsen Varberg, Sweden. The funders had no influence over study design, data collection, analysis, or preparation of the manuscript.
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
References
Bach A.G.
Schmoll H.J.
Beckel C.
et al.
Pulmonary embolism in oncologic patients: frequency and embolus burden of symptomatic and unsuspected events.
Risk of recurrent venous thromboembolism and major hemorrhage in cancer-associated incidental pulmonary embolism among treated and untreated patients: a pooled analysis of 926 patients.
Incidental pulmonary embolism in patients with cancer: prevalence, underdiagnosis and evaluation of an AI algorithm for automatic detection of pulmonary embolism.
00049-X&id=ga1.jpg){kind=link}
00049-X&id=gr1.jpg){kind=link}
00049-X&id=gr2.jpg){kind=link}
00049-X&id=gr3.jpg){kind=link}
00049-X&id=gr4.jpg){kind=link}