Evaluating growth patterns and rupture risk to determine the repair threshold for saccular abdominal aortic aneurysms

Saccular abdominal aortic aneurysms (sAAAs) lack a specific International Classification of Diseases code, and data on clinical outcomes and size criteria for repair remain sparse.

We performed a large language model (LLM)-assisted term search to identify all consecutive patients with sAAA at a large multisystem hospital between April 1999 and August 2025. Patients with sAAA (group I) were compared with patients with fusiform AAA (fAAA; group II) using deidentified LLM data in the entire cohort. Clinical data including demographics, mode of detection, aneurysm size, and outcomes were retrospectively reviewed from a single center. The primary outcomes were rupture risk and aneurysm growth rate.

The LLM accessed 1.3 billion clinical notes from 8 million patients over the study period and identified 298 patients (233 male; mean age, 75 ± 8 years) with sAAA and 42,971 with fAAA. The mean fAAA diameter was 4 ± 1 cm; no sAAA ruptures were observed at presentation or during long-term follow-up, regardless of aneurysm size. Clinical data validation was performed in 50 patients with sAAA (33 male; mean age, 81 ± 9 years; median initial maximum diameter, 4.1 cm [range, 3-6.2 cm]). The majority were detected incidentally. Thirty-two patients (64%) did not undergo repair. In this group, the initial mean sAAA diameter was 3.6 ± 1.0 cm. At a mean imaging follow-up of 57 months (average 3.5 scan/patient), the mean sAAA diameter increased to 4 ± 1.3 cm. The rate of aneurysm growth increased with increasing sAAA size (P = .03). Eighteen patients underwent repair (mean sAAA diameter, 5.4 ± 1.0 cm). Repair was performed because sAAAs were symptomatic in three patients, meeting traditional size criteria (5 cm in female and 5.5 cm in male patients) at the initial scan in two patients or on surveillance (mean, 32 months) in 12 patients. Repair was based on morphology alone in one asymptomatic patient. Procedures included open repair in four and endovascular repair in 14 patients; operative mortality was 2%. At a mean clinical follow-up of 80 months (median, 64 months), there was no additional aneurysm-related mortality.

Rupture of AAA due to saccular morphology seems rare. Most sAAAs present incidentally and demonstrate slow growth rates when <5 cm in diameter. Intervention is safe and effective for sAAAs that meet the traditional criteria for repair for AAA. LLMs demonstrate potential utility in the identification of rare vascular pathologies that do not have defined International Classification of Diseases codes.