14484
clinical study

Natural History of Untreated Acute Pulmonary Embolism in Patients With Cancer: A Retrospective Cohort of Missed Pulmonary Emboli

Materials & Methods
This retrospective study evaluated adult cancer patients undergoing chest CT with IV contrast between Jan. and Dec. 2019 at Mayo Clinic. A commercially available AI algorithm (Aidoc, Tel Aviv, Israel) was used to identify incidental pulmonary emboli (iPE) while a natural language processing (NLP) tool analyzed radiology reports to flag unreported PE. Cases positive by AI but negative by NLP underwent manual review by a thoracic radiologist to confirm missed iPE. Confirmed cases were followed for recurrent venous thromboembolism (VTE), bleeding and mortality.

Results
Among 14,453 CT scans, 254 iPE were identified (1.7%), including 45 missed cases (0.31%). After exclusions, 33 untreated cancer patients with confirmed missed PE were analyzed. Recurrent VTE occurred in 30.3% of patients (rate 19.1 per 100 person-years), while clinically relevant non-major bleeding occurred in 18.2%. No fatal PE or major bleeding was observed, though overall mortality reached 39.4%, higher in subsegmental versus segmental PE (p = 0.049).

Conclusions
Missed and untreated pulmonary emboli in cancer patients are associated with high VTE recurrence and significant bleeding risk, even in small or subsegmental clots. These findings support guideline recommendations to consider anticoagulation for incidental or asymptomatic PE in malignancy, emphasizing the importance of early detection through AI-enabled imaging review.

Explore the Latest AI Insights, Trends and Research