3D Quantified Morphologic Predictors of Intracranial Aneurysm Instability: A Longitudinal Study.

Background and purpose: The performance of the current prediction models for intracranial aneurysm growth and rupture is suboptimal, and new markers are needed to improve prediction. There is a strong need for longitudinal studies that use standardized morphologic parameters. In this longitudinal study, we aimed to identify standardized 3D quantified morphologic parameters as predictors of aneurysm growth or rupture during long-term follow-up.

Materials and methods: We used a database of consecutive patients with saccular unruptured intracranial aneurysms diagnosed between 2008 and 2018. Using a retrospective case-cohort design, we included a computer-generated random sample of aneurysms from the full cohort and aneurysms with growth or rupture during follow-up outside the random sample. The case-cohort design is efficient for low-incidence outcomes while maintaining the temporal association between exposure and outcome. Aneurysms were annotated on baseline CTA or MRA images, and 3D morphologic parameters were quantified. Univariable and multivariable Cox proportional hazards models were used to identify 3D morphologic predictors of either aneurysm growth or rupture. An inverse sampling probability weight was applied to obtain unbiased estimates of the hazard ratios.

Results: We included 278 patients (median age, 59 years [interquartile range, 50-66]; 209 women) with 327 aneurysms, of which 239 aneurysms were stable during follow-up (73%), 68 grew without subsequent rupture (21%), 7 grew with subsequent rupture (2%), and 13 ruptured without preceding growth (4%). In the multivariable model for growth prediction (median follow-up, 4.1 years [interquartile range, 1.9-7.1]), major axis (hazard ratio, 1.16; 95% CI, 0.84-1.61) and shape index (hazard ratio, 1.53; 95% CI, 0.76-3.08) were retained, with a c-statistic of 0.56 (95% CI, 0.49-0.63). The same parameters were retained in the multivariable model for prediction of rupture (median follow-up, 4.5 years [interquartile range, 2.1-7.3]): major axis (hazard ratio, 2.27; 95% CI, 1.36-3.80) and shape index (hazard ratio, 3.33; 95% CI, 0.95-11.62), with a c-statistic of 0.85 (95% CI, 0.77-0.94).

Conclusions: We identified major axis and shape index as candidate 3D-quantified morphologic predictors of both aneurysm growth and rupture, but only for rupture did they demonstrate good discriminative power in our cohort. These parameters will need external validation and should be integrated with existing clinical prediction models.