Date of Degree
PhD (Doctor of Philosophy)
Madhavan L. Raghavan
Most intracranial aneurysms (IA) that are unruptured during clinical screening remain stable over time with no measurable change or symptoms if left untreated. However, a few do grow larger and may rupture. The ability to preemptively identify aneurysms that will become unstable over time (i.e., those that will grow and/or rupture) can result in timely intervention for these few patients while avoiding unnecessary treatment for countless others. So far, reports have primarily assessed potential rupture factors in IAs between ruptured lesions and unruptured lesions. These factors (that discriminate rupture status) do not necessarily distinguish unruptured aneurysms that tend towards growth and/or rupture over a period of time from those that remain stable. The hypothesis motivating this study is that aneurysm shape morphology provides critical information that will indicate instability in unruptured aneurysms. We tested this hypothesis in a large population multi-center prospective longitudinal cohort study of unruptured IAs.
A total of 198 study subjects were recruited at four clinical centers - Penn State Hershey Medical Center, Massachusetts General Hospital, Thomas Jefferson University Hospital and University of Iowa Hospitals and Clinics. All had at least one unruptured IA and were placed on follow-up without immediate intervention. Three-dimensional volumetric scans were obtained from the clinics. Three dimensional models were created from the source data using levelset segmentation techniques implemented in the Vascular Modeling ToolKit (VMTK). Five size and six shape indices were calculated on the isolated aneurysm geometry: Size indices - Height (H), max. diameter (Dmax), neck diameter (Dn), volume (V), and surface area (SA) and Shape Indices - Undulation index (UI), Aspect ratio (AR), Ellipticity index (EI), Non-sphericity index (NSI), Bulge Location (BL), and Bottleneck factor (BF). Pressure-induced wall tension was calculated using Finite element analysis. An isotropic, polynomial model was used after a comparative study of different modeling choices was performed and the 95th percentile max. principal stress (Peak Wall Stress - PWS) was chosen to be the index of maximum wall tension. Independent studies were performed to document validity of methods used, to quantify sensitivity of these indices (in turn, of the segmented geometries) to imaging modality and to quantify variability in segmented geometries between investigators. Researchers were blinded from the follow-up status of IAs to avoid any user bias in the analysis. During follow-up, aneurysms were labeled as "grown" or "stable" by radiologists. Study subjects were assigned to one of two groups based on clinical outcome: Unstable group (grown or ruptured) or Stable group (remained same or lesser size). Statistical analysis was then performed on the computed indices. One hundred ninety eight aneurysms were followed for an average of 607 days. Over the follow-up period, 26 were lost to follow up, 150 were found to be stable, 20 had grown, 1 deceased, 1 could not be located on the angiogram and none had ruptured. Also, of the total 198, 53 were electively treated during follow-up. Non-parametric Mann-Whitney U-tests were performed for index to test for statistically significant difference between the stable (n1 = 150) and the unstable group (n2 = 21). None of the indices differentiated the two groups in a statistically significant manner. Receiver Operating Characteristics (ROC) curve analysis was consistent with these findings. Is aneurysm morphology (size and shape) a risk factor for IA growth/rupture? We sought to test this hypothesis in a study design that is unique in many ways - it is the first large population study of longitudinal cohorts; the study population was prospectively recruited at multiple clinical sites; the study population was predominated by small aneurysms (88% of IAs in the study population were less than 7 mm). The results of this large population cohort study suggest that previously reported metrics of aneurysm cannot serve as prognostic indicators of longitudinal outcome.
xii, 140 pages
Includes bibliographical references (pages 105-110).
Copyright 2012 Manasi Ramachandran